Skip to content

The rule gets paper money. The S&P 500 sets the bar.

Paper strategies get $1,000 of paper money at real prices, then are measured against just buying the S&P 500 over the same dates. Observed wallets and claim checks keep their own ruler, explained below.

One rule. Two uses for the same money.

Start both clocks on the strategy’s own dates. Compare returns, then subtract the S&P 500’s return in percentage points. Equal paper capital is a paper-strategy rule: observed wallets and claim hit rates are not compared with the index.

Publish the rule before the result.

The original source, the entry, the stake and the exit stay on the strategy page. Paper trades use real market prices and fees.

Follow the strategy

$1,000.00 → $1,229.13

Rotten Tomatoes score
+22.9% +$229.13 since its first trade on 24 Oct 2026

Just buy the S&P 500

$1,000.00 → $989.20

SPY, 23 Oct 2026 – 16 Nov 2026
−1.1% over the same comparison window

+24.0 pts vs S&P

The strategy is ahead of the index over these dates. A short record is a forward test, not proof of an edge.

Real worked example · Mon 16 Nov, 09:25 UTC snapshot. Both branches start with $1,000.00 of paper money over the same comparison window. Equal paper capital is a paper-bot rule, not a statement about observed wallets or claim hit rates. Read the strategy’s full record →

What this board tests

Four kinds of strategies at real prices: Trading: viral prediction-market systems, crypto models and AI agents. Sports: our own models (who wins, and over/under totals), viral betting systems and claim checks. Wallets: copying the 'smart money' wallets, plus observed wallets: public real-wallet evidence. Investing: the boring baseline: the S&P 500, Berkshire's portfolio and the wheel. Paper strategies each get $1,000 of paper money.

The bar: just buying the S&P 500

One bot puts its $1,000 into SPY (the S&P 500 fund) and never sells. Paper strategies are measured against it over the same dates: "vs S&P" is the strategy's return minus what SPY did since that strategy started, in percentage points (+2.0 pts = 2% better than just buying the index). Observed wallets show public real-wallet evidence, and claim checks show the picks and hit rate against break-even; both use their own yardstick. The question the board asks: does anything beat investing carefully?

Some records need a different ruler.

Real wallets are someone else’s on-chain trading history. One isn’t our paper strategy: we don’t hand it $1,000 or grade its dollars against the S&P 500. Copying a wallet is a separate paper strategy with its own starting money and dates.

Claim checks publish picks and measure the hit rate against break-even. Without free sportsbook parlay prices there is no honest paper P&L to compare with the S&P 500, so the hit rate stands on its own.

Read a wallet record →Read a claim check →

The scorecard

P&L
Profit or loss in dollars: the paper account’s current value minus its starting money.
Return
The paper account’s gain or loss as a percentage of its starting money. Record facts show the lifetime return; chart ranges show their selected dates.
vs S&P 500
The strategy’s return minus the S&P 500’s return over the same dates, in percentage points. A 5% return minus a 3% return is +2.0 points.
Max drawdown
The largest fall from an earlier peak in recorded account value. The facts show the lifetime maximum; the band shows drawdown during the selected chart dates.
Win rate
For settled prediction-market bets, the share that won. It excludes open bets and does not show how much was gained or lost.
CLV (closing-line value)
The market’s price just before close minus the entry price, averaged over settled bets. Positive CLV means the entry price was lower. It does not by itself establish skill, especially in a small sample.

Markets and prices

Polymarket
Polymarket sells yes/no shares on real events. A share pays $1 if it comes true and $0 if not.
Market price
A share at 42¢ means the crowd gives it about a 42% chance.
Buying a share
Buy at 42¢ and win: you get $1, a 58¢ profit per share. Lose: the 42¢ is gone.
Fees
Polymarket charges a small fee when you buy (about 1¢ per share on a 50/50 market, less near 0¢ or 100¢).
Edge
How much better the bot thinks its price is than the crowd's, after fees.

Why the house bots skip most games

Our sports models publish a pick for every game, so their public record stays honest. Betting every pick leads nowhere, so each house bot follows a tested policy and bets only a few: Edge window: the model must beat the price by enough (say 3%), but not by too much. A huge edge usually means the model is wrong, not the market. Payout band: no heavy favourites that pay pennies and no longshots; totals only near even money. Markets that worked: only the markets and sides that held up in backtests on seasons the policy never saw. Caps: at most a few bets a day or week, best edge first. On models page every pick shows Bet or Pass with the reason. "Off" means the backtest found no edge worth taking: the bot places no bets and the picks are still graded. A bot with no policy yet keeps the old rule (any edge of 4¢ or more).

Execution and timing

Paper execution
Paper strategies play with $1,000 of fake ("paper") money at real prices, with real fees. TheTapeTest places no real orders.
Check cadence
Most bots check Polymarket once an hour; the retired RohOnChain AI agents checked every 15 minutes. We tested whether checking every minute would help the sports models, on 262 NFL games from 2025. It didn't: they placed nearly the same bets at the same prices, because a model's disagreement with the market lasts days, not minutes. What helps is publishing picks early.

The fine print

Fills: a buy walks the real order book (cheapest offers first) until the stake is spent or limits are hit. Partial fills only on Nothing Ever Happens, Favourite bonds, and Longshot lottery; Weather, Weather wallet replica, Mention NO, house models, Copy the leader, and Bitcoin daily ladder require a full stake. Fees: shares × fee rate × price × (1 − price), only on markets with fees enabled (sports: 5% rate, so about 1¢ per share at 50¢). Settlements are free. CLV: closing price from token history just before the market closed minus our entry price, averaged over settled bets. Open bets are excluded. AI: the RohOnChain desk and the Jev launch desk were retired on 28 Oct 2026. No model is called anymore: the board runs on plain code and every paper trade is still public. Paper vs wallet: paper bots run in our database against public market data. The @hightemptation row is an on-chain Polymarket wallet; its P&L comes from the public Data API, not the paper engine. Public trade feed: bot pages call GET /bots/{slug}/trades with no delay. Each page shows the equity history, settled and open bets, and the latest 100 raw fills; older fills are left out.

Bring a rule worth testing.

A source, an entry, an exit. I review every source before any paper money goes behind it; credit stays with whoever posted the original rules, and a listing isn’t automatic. The losing record stays public too.

Submit a strategy →