September 26, 2026 · 22 min · desk
How to use Agents61 to make money: a desk tutorial
Eight gates before capital moves
Read this top to bottom. A later gate does not repair an earlier stop. The paper percentages further down are a reading drill from 26 September 2026, not a forecast.
- 1
What is the asset?
Pass. A US stock goes to Analyze, Isolated. A coin goes to the blockchain desk. A yes/no event goes to Predictions.
Stop. You force a stock multiple onto bitcoin, or you treat an event contract like a share. Stop. Wrong file.
- 2
Do you already have one name?
Pass. Isolated. Each seat writes alone. You are here for the split, not for a blended score.
Stop. The question is still “where might capital flow?” Use division of labor as a pipeline, then pick one name. Do not flip the method mid-run.
- 3
Can you write the kill in one sentence?
Pass. A fact that would make you sell, restated with the page closed. Example: services growth stalls and the multiple stays elevated while buybacks slow.
Stop. You are still hunting a rating. Do not open a paper book for comfort, and do not open a broker.
- 4
Which book matches the account you have?
Pass. Cash, long-only: Shares. A financed long/short you can actually monitor: Contract. Cash to be assigned, plus options vocabulary: Wheel.
Stop. You pick the style with the highest ending percent. That number is not a job. See the AAPL row: Shares +68%, Wheel +6% with a worse hole.
- 5
What window is on the card?
Pass. You can name the dates. These tables: crypto dailies 2024-01-01 to 2026-09-26; US dailies 2022-09-30 to 2026-09-25. One thousand bars.
Stop. You treat one regime as the next regime. The US window starts near a bear-market low. That inflates long-only endings.
- 6
What is the hole in your dollars?
Pass. Max drawdown times your account, not the virtual $10,000. You can sit through that hole without turning the rule off.
Stop. You would have quit halfway. The ending return is not yours. AAPL shares −13.77% is about $275 on $2,000 and about $6,900 on $50,000.
- 7
Do win rate and profit factor agree?
Pass. You can say both. Profit factor is closed-trade gains divided by closed-trade losses. Near 1 means the wins do not pay for the losses.
Stop. You quote the return alone. SPY wheel wins 74% of closes and still has a profit factor of 1.05. AAPL wheel is +5.81% with a profit factor of 0.40.
- 8
Is the live gate actually open?
Pass. Kill written, hole affordable, style matches the account, size cannot change rent. Then you place the order at your broker, with the exit the same day.
Stop. Any line missing. Agents61 still does not send the order. A missing line means the money stays put.
| Move | You leave with | You do not leave with | Money logic |
|---|---|---|---|
| Isolated split | A kill sentence | A buy rating | Refusing a bad idea is the first P&L |
| Shares / Contract / Wheel | Drawdown and a trade list | Income | The curve is a rehearsal on $10,000 of paper |
| Blockchain perpetual | A daily rule on public candles | A Binance fill | No liquidation engine, no USDT movement |
| Predictions | A probability next to a story | A wallet bet | If your story cannot beat a stated mid, it is a mood |
| Quant Lab script | A rule you can read | A curve fitted after the fact | Editing the script to flatter history is not evidence |
| Paid seating | More isolated writers | A higher win rate | You pay for seats that write, not for luck |
Observer is $0 and can already open the public US quant and blockchain books. Seating changes how many prompts run.
| Book | Window | Return | Max hole | Closed | Win | Profit factor | Judgment |
|---|---|---|---|---|---|---|---|
| BTC contract | 2024-01-01 → 2026-09-26 | +5.95% | −6.76% | 25 | 44% | 1.53 | About +$595 after a $676 hole. Not a salary. |
| ETH contract | same crypto window | +6.94% | −5.27% | 23 | 52% | 1.89 | Cleaner tape. Do not average it with a value seat that rejects coins. |
| AAPL contract | 2022-09-30 → 2026-09-25 | +1.09% | −8.85% | 26 | 35% | 1.01 | About +$109 after an $885 hole. Fail. |
| AAPL shares | same US window | +68.39% | −13.77% | 14 | 57% | 3.11 | The pretty row. Only yours if you held −13.77% from a 2022 low. |
| AAPL wheel | same US window | +5.81% | −21.66% | 53 | 53% | 0.40 | Curve up, closed losses larger than wins, deepest hole. Fail unless you can be assigned. |
| SPY shares | same US window | +27.55% | −11.11% | 20 | 45% | 2.47 | Fewer trades. Losses smaller than wins. Still a window, not a promise. |
| SPY wheel | same US window | +33.43% | −8.61% | 47 | 74% | 1.05 | High win rate, almost no edge per loss. Do not call 74% a machine. |
Return is the equity curve, including open marks. Profit factor uses closed trades only, so a positive return can sit next to a profit factor under 1. Crypto source: Binance daily. US source: Sina daily. Option prices are a Black-Scholes model on the stock’s own realized volatility, not an exchange chain. Re-open the book before you treat a cell as current.
Agents61.com is a research desk, not a brokerage, not a signal chat, and not a machine that deposits profit into your account. The honest way to use it to make money is narrower than the ads in this category usually claim. You use the committee to kill weak ideas before they become positions. You use the US quant books and the blockchain paper perpetual to see how a written rule would have behaved on daily bars, including the losing stretches. You use the prediction desk to separate a probability from a story. Then you decide, yourself, whether any of that deserves capital at a broker you control. The site will not click buy for you. There is no order button on the report, on the crypto board, or on the US quant tape. That limit is the product. A desk that also routed orders would be a different company, with a different license, and it would be much easier to confuse a simulation with a paycheck.
Start by separating two piles of money. The first pile is the subscription, if you pay one. Observer is free and already lets you read the public pages, including the US quant book at /research/us-quant and the blockchain board at /research/crypto. Paid seating — Analyst, Associate, Principal, Committee, Partners — buys more isolated writers on a report, not a higher win rate. The second pile is the capital you might later place at a broker. Agents61 never touches that second pile. A paper book starts at a virtual ten thousand dollars so the percentages are readable. A plus sign on that book is not income. A minus sign is not a loss in your bank account. It is a rehearsal. People who skip that distinction treat a backtest like a salary and then size the live trade as if the rehearsal were already cash. That is the fastest way to lose money while staring at a research site.
The modules you will actually touch are these. The clerk, on the dashboard, is the front door: you type a ticker or a job and it routes you. Analyze is the committee thread for one name. Discover is the opportunity board, a map of themes rather than a ranked buy list. The blockchain desk and /research/crypto are the on-chain board plus a paper perpetual on liquid crypto pairs. US quant, at /research/us-quant and on the dashboard as its own entry, is a separate book for US names: a long-short contract simulation, a long-only share book, and an options wheel that uses model premiums rather than a live options chain. Quant Lab is the workshop that compiles a Python backtest from the masters you pick. It does not run the US or crypto tape for you. Predictions is a paper desk for event contracts. Watchlist is where names you refuse to forget can sit. Private is a one-off name that is not in the public universe. Lineup is a screen across names. If a tutorial online tells you there is an eighth module that wires Alpaca, Interactive Brokers, or Binance orders, it is describing a product that is not this one.
A newcomer should not open every door on day one. The useful first hour is small. Register, land on the dashboard, and leave the method on Isolated. Type one US ticker you already know how to spell, something with filings, not a ticker you saw in a screenshot that morning. Read the seats that disagree. Write down, in your own notes, the one fact that would make you sell. Only after that sentence exists should you open US quant and load the same symbol on the Shares book. You are checking whether a simple trend rule would have been a comfortable ride, not whether the site has anointed the stock. If you cannot explain the rule in one sentence — rising medium-term trend, enter on a break of the recent high, exit on a break of the recent low or a volatility stop — you are not ready to care about the percentage at the top of the card.
How a row becomes a decision
Take AAPL on the US window, virtual $10,000. Contract made about $109 and asked you to sit an $885 hole, with a profit factor of 1.01. That is a coin-flip after costs. Fail gate 7. The wheel ended about $581 higher and asked for a $2,166 hole, while closed losses were larger than closed wins (profit factor 0.40). Fail gates 4 and 7 unless being assigned the shares is the account you already wanted. Shares ended about $6,839 higher, profit factor 3.11, fourteen closes, and a $1,377 hole. It passes the tape test and still fails gate 6 if you cannot hold a 13.77% drawdown on the account you actually have. The window also starts on 30 September 2022, close to a broad low, so the long-only ending is flattered. BTC is a different file: about $595 of paper profit against a $676 hole over the crypto window. Gate 1 keeps it off the stock checklist. Gate 8 stays shut until the kill sentence exists. The money, if it comes, is the size of a rule you did not abandon, placed at your broker. Most rows in the table are there to be refused.
What the committee is for
The committee exists to stop you from falling in love with the first fluent paragraph. Isolated mode is the default and the one you should use when you already have a name. Each unlocked seat sees the ticker and its own rules. Buffett does not read Wood. Burry does not read Lynch. A clerk stacks agreements and splits after the notes exist. Division of labor is the other switch, and it is a pipeline rather than a group chat: trend can hand a map to cycle, cycle can hand a regime note to selection, selection can hand a pool to the red team. Peers in the same group still do not see each other. Use Isolated for “what is wrong with this name.” Use Division of labor when the question is a sequence — where capital may be flowing, where the cycle sits, which names survive — and you are willing to read handoffs instead of a single verdict. Neither switch prints a buy rating. If the page feels unfinished because it refuses to say “buy,” it is working.
How to read a split instead of hunting a rating
A useful report is a split you can act on, not a score you can obey. When quality seats and red-team seats disagree, leave the disagreement on the page. Your job is to decide which look-at list matches the risk you can actually hold. A value seat asking about owner earnings is not “wrong” because a growth seat is excited about duration. They ran different tests. Write three lines before you leave the report: what would have to be true for the bull case, what would kill it, and what you will do if the kill arrives. That third line is the part most people skip, and it is the part that protects money. Agents61 will not size the position for your life. Kelly language from a quant seat is a research overlay, not a prescription for your rent. If filings are thin, a serious seat should stay inconclusive. Treat a confident paragraph with no numbers as a draft, not as permission.
Seating, cost, and what you are actually buying
Seating changes how many isolated prompts run, not how lucky the model is. The working set starts at Analyst, sixteen seats that already cover era, cycle, selection, a red team, timing, and sizing. Associate adds red-team, cycle, and timing seats. Principal, Committee, and Partners light more of the roster, including exit discipline at the top end. Partners is the same full roster as Committee with more room to run, not a secret strategy. Empty seats stay empty on purpose. Paying for a louder chat would be cheaper and worse. Observer at zero dollars is enough to learn the public books and to read this tutorial against the live pages. Upgrade when you are running the same name often enough that missing seats annoy you, not because a pricing card implied a return. If the subscription costs more than you are willing to lose on a research hobby, you are already sizing the wrong pile of money.
The opportunity board is a map, not a shopping list
Discover, the opportunity board, is where you go when you do not yet have a ticker. It is organized as themes and cycles, the kind of “where might capital flow” question that a single-name report cannot answer. Use it to pick a neighborhood, then take one name from that neighborhood into Analyze. Do not treat the board as a list of things to purchase on Monday. A theme can be early, crowded, or already priced. The board does not know your taxes, your job, or whether you will panic at a fifteen percent drawdown. The money-making move at this step is subtraction: cross off themes you do not understand well enough to hold through a dull year. A smaller map that you can explain is worth more than a larger map you will abandon.
US quant is three paper books, and none of them trade
The US quant module is separate from the blockchain desk. It covers liquid US names and index funds on the page: SPY, QQQ, IWM, DIA, AAPL, MSFT, NVDA, AMZN, GOOGL, META, and TSLA. Daily bars come from a public equity history feed. The simulation starts at ten thousand virtual dollars. There are three styles, and mixing them up is how people misread the tape. Contract is a long-short research book on the stock’s own daily bars. It is not a listed futures contract and not a CME product. It applies a financing charge while long and a borrow charge while short so the short side is not free. Shares is a long-only cash book: no short, a small commission, the same trend entry, and an exit when the channel breaks or a volatility stop hits. Options wheel is a third book: while price holds above a rising medium-term average it sells a modeled cash-secured put; if that put finishes in the money the book is assigned shares and then sells covered calls until called away or stopped. Premiums are Black-Scholes prices on the stock’s own recent realized volatility. They are not quotes from an options exchange. If your broker’s chain disagrees with the model, the broker is the price you would actually receive. The model is there so you can see the shape of a wheel — assignment, calls, expiries, stops — before you learn that shape with real premium.
How to use a paper tape before you size real capital
Open /research/us-quant or the US quant entry on the dashboard. Pick the symbol you already wrote a kill-condition for. Read the style label so you know whether you are looking at Contract, Shares, or Options wheel. Then read the stats in this order, and refuse to stop at the return. First, the window: which dates the daily bars cover. A pretty return on a window that was one regime is a story about that regime. Second, max drawdown: the worst peak-to-trough on the equity curve. If you would have turned the book off halfway through that hole, you cannot claim the ending return. Third, the trade count and the win rate. A high win rate with a profit factor near one means the wins are small relative to the losses. Fourth, the open line, if there is one: a backtest that is still in a position is not a finished experiment. Fifth, the trades table. Each row has a side, an instrument label on the wheel (shares, put, or call), entry and exit, a reason, and a result. Reasons you should be able to translate: channel, stop, assigned, called, expiry. If you cannot say why the last closed trade ended, you are looking at a scoreboard, not a strategy.
Contract, shares, and the wheel are different jobs
Use the Contract book when the question is whether a long-short trend rule survived costs on that name. Shorts are allowed. Financing is not crypto funding; it is a simple equity carry so leverage does not look free. Use the Shares book when the question is the one most people actually face: long-only stock, cash, no options vocabulary. It will trade less often than the contract book because it cannot profit from a downtrend except by being flat. A stretch of cash is a result, not a bug. Use the wheel when you want to study premium selling on the same daily bars: roughly thirty-delta short puts, about a month of trading days, assignment into shares, then covered calls. The wheel’s drawdown often shows up as share inventory you did not emotionally plan to own. That is the lesson. A wheel that only looks good while puts expire worthless has not been tested. Switch styles on the same symbol and write down which path you could actually follow. The path you could follow is the only one that can make you money. The path with the prettier ending that you would have abandoned is a chart.
Blockchain research uses a different book
Crypto does not belong inside the US share book, and the site keeps it on its own desk for that reason. The blockchain page and /research/crypto cover a paper perpetual on liquid pairs: BTC, ETH, SOL, BNB, XRP, DOGE, ADA, AVAX, LINK, TON, SUI, and LTC. Daily candles prefer Binance public history. The rule is the same family as the equity contract — a Donchian channel for entry and exit, a rising or falling medium-term average as a filter, an average-true-range stop, a volatility gate that skips dead or chaotic markets, a cap on notional, a fee each side, and a cooldown after a stop so the book does not immediately re-enter the same direction. Funding is a research estimate from how far price sits from its medium-term average, clipped so funding cannot silently become the whole result. Longs pay when price is stretched above that average. This is not a live Binance futures account. It will not move USDT. Read it the same way you read the US contract book: drawdown first, then whether you understand the exits. Value seats on the committee will often stay skeptical of coins with no cash flow. That skepticism is allowed to remain next to a trend book that traded the coin anyway. Do not average them into a single coin rating.
Quant Lab is a workshop, not the tape
Quant Lab, on the dashboard, lets you pick masters and compile a Python backtest you can read. It is the place for a custom question: a different indicator, a different holding rule, a script you want to inspect. It does not replace the US quant tape or the crypto perpetual tape. Those two already run a fixed rule and show trades. If you export a script, you are leaving the site’s numbers behind and entering your own. That is useful, and it is also where people quietly change a parameter until the curve looks like income. If you do that, label the result as fitted. A rule that was edited after seeing the same history it is judged on is not evidence you can spend. The built-in books are rigid on purpose. Their job is to be the same rule tomorrow as today, so a change in the curve comes from new bars, not from a silent edit.
Prediction markets are a probability desk
The predictions module studies event contracts — questions that resolve yes or no — as research, not as a betting slip with your credit card attached. Use it when the committee is arguing about an event that has a public probability somewhere, and you want to see whether your story is more confident than the market. A story that cannot beat a stated probability is often just a mood. Paper fills, when the desk shows them, are a rehearsal of paying the spread and being wrong in public. They are not a bankroll. Do not move rent money onto a prediction venue because a research page framed a question cleanly. The money-making use of this module is refusal: you decline a stock story whose key event is priced at a probability you had not bothered to look up.
Watchlist, private names, and the lineup
After a report, put the name on the watchlist if you still care and you are not ready to act. The watchlist is a memory, not a trigger. Private is for a name outside the public universe: a ticker the board does not list, a situation you do not want mixed into the theme map. You still get research, not an order. Lineup is the screen you run when you want the same question asked across several names, under the method you already picked. Run it after you understand one name deeply. A screen of twenty tickers you have not inverted is a way to feel busy. One name with a written exit, plus a paper book you have actually read, is a way to get paid later for being selective now.
A four-week operating rhythm
Week one is orientation. Read this note. Open the homepage guide. Convene one US ticker in Isolated mode. Write the kill-condition. Open the Shares book for that ticker and read the drawdown and the last ten trades. Do not place a live order this week. Week two is comparison. Run the same ticker on Contract and, if you understand options vocabulary, on the wheel. Write which book matches the account you actually have. Most brokerage accounts are closer to Shares than to a financed long-short book. Week three is the neighborhood. Use Discover or a lineup to find a second name in the same theme, and run the committee again. If you cannot explain why the second name is different from the first, you do not own a theme, you own a duplicate. Week four is crypto or a prediction, not both as a distraction. If you care about BTC or ETH, use the blockchain perpetual and keep the equity committee’s skepticism visible. If you care about an event, use the prediction desk and write the probability you believed before you looked. At the end of the month you should have two written kill-conditions, one paper book you understand, and a list of ideas you refused. The refusals are the part that compounds.
Where money actually moves
When you eventually place capital, it moves at your broker, under your name, with your tax lot, your pattern-day rules, and your margin agreement. Agents61 will not know that you clicked. A practical bridge looks like this. You only consider a live share purchase after three things are true: the committee split is something you can restate without the page in front of you, the Shares book drawdown is a loss you have already imagined in dollars on your real account size, and the kill-condition is an order you could enter the same day you enter the shares. Scale from a size that cannot change your rent. The paper book’s ten thousand dollars is a unit for reading percentages, not a suggestion that you start there. If your live account is two thousand dollars, a twenty percent paper drawdown is four hundred dollars, and you should picture that number, not the percentage. Options are a later step. A wheel uses cash to secure the put and then owns shares. If you do not have the cash to be assigned, you are not running the wheel the book studied. You are running a different, more fragile trade. Crypto perps at a real exchange add liquidation, funding that is not our estimate, and a counterparty. Nothing on the blockchain page prepares you for a liquidation engine. If you trade one, you are leaving the tutorial.
Records worth keeping
Keep a plain log outside the site. Date, ticker, the method you used, the split in one sentence, the paper book and its drawdown, the kill-condition, and — only if you acted — the live size and the live exit rule. Review it monthly. The point of the log is to catch the moment you start overriding your own exits because a new paragraph sounded smarter than the old one. Also record ideas you refused, and why. A refusal log is how you learn whether the desk is saving you money. If every refusal would have been a winner over the next month, your kill-conditions are too trigger-happy, or you are only writing down the refusals that hurt. If every refusal would have been a loser, you are finally using the red team as something other than decoration. Neither result shows up in a single backtest statistic.
Failure modes that cost people money
The expensive mistakes are repetitive. Treating the paper return as expected income. Averaging Buffett and Wood into one score because a split feels untidy. Forcing a price-to-earnings story onto bitcoin. Switching the US book to the wheel because the ending number was prettier, without being willing to own the shares. Re-optimizing Quant Lab until the curve flatters the same history. Raising seat count mid-argument because you want the room to agree. Moving from a research site to a high-leverage perp because the paper perpetual “worked” on a calm sample. Ignoring the source line when a feed fails and the page marks the series as a sample path: a sample path is a labeled rehearsal, not market history, and it must not be used as evidence. Ignoring fees, funding, and carry because they are small on one trade and large across a year. And the oldest one: raising size after a winning paper stretch and cutting size after the drawdown you were supposed to have planned for. The site can show you the drawdown. It cannot make you sit through the live one.
A worked pass on one name
Suppose the name is a mega-cap you already hold in your head, AAPL. You open the dashboard, Isolated, and convene AAPL. You read a quality seat and a red-team seat and you write one kill: you would be wrong if the services mix stopped growing and the multiple stayed elevated while buybacks slowed. You do not need the site to bless that sentence. You open US quant, select AAPL, and start on Shares. You note the date window, the drawdown, and whether the trades are infrequent trend rides or a choppy mess. You switch to Contract only to see whether shorts would have saved you in a decline you remember living through. You switch to the wheel only if assignment would fit your cash. You do not pick the style with the highest ending equity. You pick the style whose worst hole you can describe in dollars. Then you do nothing live until the kill-condition and the size are both written. If a week later the committee note and your log disagree, you re-read before you add. That loop — convene, invert, paper the matching book, size only what the drawdown allows — is the whole earning method this site can support. Everything else is entertainment.
What to ignore while you learn
Ignore anyone who posts a screenshot of a paper equity curve as proof the desk pays. Ignore composite ratings, including ones you are tempted to invent from the clerk’s stack. Ignore intraday noise; these books are daily, and a daily rule cannot justify a five-minute trade. Ignore the urge to add more pairs before you understand one. The crypto list and the US list are long enough to waste a month. BTC or SPY is a better first chart than a name you cannot describe. Ignore seating as a status symbol. Sixteen serious seats that disagree are more valuable than sixty-one seats you will not read. And ignore the hope that a later version will add a buy button and remove the need for judgment. If that button ever appeared, the tutorial would have to shrink to a warning. Judgment is the part you are paid for, if you are paid at all.
A newcomer’s first sitting
If you are opening the site for the first time, use the guide on the homepage and the same guide at the top of the dashboard clerk. It is the short version of this note: one ticker, Isolated, read the split, then the matching paper book, and only then a live decision somewhere else. The public US quant page and the public crypto page do not require a paid seat, so you can verify the tapes before you subscribe. The committee’s depth is what seating changes. Come back to this essay when a paper curve tempts you to skip the kill-condition. The curve will still be there after you have written the sentence. Your capital should not move until the sentence exists.
FAQ
- Does Agents61 place trades or send money to my account?
- No. Reports, the US quant books, the blockchain perpetual, Quant Lab, and the prediction desk are research simulations. There is no buy button and no broker connection. Any live order happens at a broker you choose, under your own name.
- Can I treat the backtest return as expected profit?
- No. The books start from a virtual ten thousand dollars on daily bars, with modeled costs. They show a path, a drawdown, and a trade list for a fixed rule. They are not a forecast, not a track record of customer accounts, and not advice.
- Which module should a newcomer open first?
- The dashboard clerk, then Analyze on one US ticker in Isolated mode. After you can write a kill-condition, open US quant on the Shares book for that same ticker. Use the blockchain desk only when the asset is actually crypto.
- Are the options prices real quotes?
- No. The wheel prices options with a Black-Scholes model on the stock’s own recent realized volatility. Your broker’s chain is the price you would actually get. The book is for studying assignment and covered calls, not for copying a premium.
- Is the crypto perpetual a live Binance account?
- No. It replays a daily rule on public candles, with a research funding estimate and fees. It does not move USDT and it is not a liquidation engine. Do not size a real perp from the paper result.
Related notes
Put the note into practice on your desk
Register to staff isolated masters on a ticker, or open Prediction Markets for probability-gap research. Same compliance line as this post — simulation, not advice.
Create accountAgents61 is a research simulation. Not investment advice.