Polytutor

Round 6, 6/39: Inventory-Skewed Adaptive MM. Quote YES and NO around a microprice, skew for inventory, widen for volatility and toxic flow. Reality: one side fills first, thin books pick you off, and queue churn kills priority. Research only. github.com/Polytutor-AI-Labs…

Round 6, 5/39: Depth-Imbalance Rebalance. Lean the way top-5 YES vs NO bid depth is stacked on a Polymarket Up/Down book, exit when it fades. Reality: size gets pulled, thin books inflate the signal, and it decays before a maker fill. Research only. github.com/Polytutor-AI-Labs…

Round 6, 4/39: BNSB. Wait for Binance to squeeze into a tight range, freeze its midpoint, then in the last 30s trade a break away from it. Reality: a squeeze doesn't pick a direction, breaks fake out, and the token is often already pricey. Research only. github.com/Polytutor-AI-Labs…

Round 6, 3/39: BNPTB. Hourly Up/Down: anchor to the hour's first Binance print, trade the side price is on, re-enter once on an opposite cross. Reality: Binance's open isn't the settlement price, so being right on Binance can still lose. Research only. github.com/Polytutor-AI-Labs…

Round 6, 2/39: BNMV. In the last 10s of a crypto Up/Down market, a big 1-second Binance move triggers a buy that way, held to settlement. Reality: a 1s spike is often exhaustion, the token is already pricey, and it can flip before a fill. Research only. github.com/Polytutor-AI-Labs…

Round 6, 1/39: BN2S. In the last ~2s of a 5-min market, check if Binance mid is above or below its opening anchor and buy UP or DOWN. Reality: by then the right side often costs 95c+, and an order can land after the window shuts. Research only. github.com/Polytutor-AI-Labs…

Round 5, 39/39: Pre-Order Market. Rest bids on the next 5-min market before it opens, hoping for queue priority while the book is thin. How it fails: your orders are older than the news, so the side that fills is often the wrong one. Research only. github.com/Polytutor-AI-Labs…

Round 5, 38/39: Live Sports State Fair Value. Turn game state (score, inning) into a win-probability model and compare its move with the market's after each play. Tell the edge is gone: the book reprices before your feed shows the play. Research only. github.com/Polytutor-AI-Labs…

Round 5, 37/39: TWAP Lock-In. Late in a 5-min Up/Down round, hold the Chainlink price forward, project the final 60s TWAP, and buy the side clearing the price to beat only if still cheap. Fails if the last seconds move or it's priced in. Research only. github.com/Polytutor-AI-Labs…

Round 5, 36/39: TWAP Cross Hunter. When the 60s settlement TWAP flips sides of the price to beat, buy the new favorite before the quote reprices. Dead-edge tell: the ask moves before your cross fires, and flip-backs eat the cents. Research only. github.com/Polytutor-AI-Labs…

Round 5, 35/39: 1¢ Buy. Rest tiny bids at 1-3¢ on short-dated outcomes and wait for a rare reversal or settle. Live-ops tell it's dead: your 1¢ fills cluster right before the outcome goes to zero, and the 1¢ size shown was never real. Research only. github.com/Polytutor-AI-Labs…

Round 5, 34/39: Low-Side Dual Reversion. When YES and NO both look cheap (~35-45¢), buy both for a sub-$1 pair, or hold one leg for reversion. Live-ops tell it's dead: one leg fills, the other runs, and 'pairs' only existed on candle lows. Research only. github.com/Polytutor-AI-Labs…

Round 5, 33/39: 99¢ Sniper. Buy the favored side near 99¢ late to collect the gap to $1. Payoff is ~1:99, so it needs >99% true odds after fees. Live-ops tell it's dead: stale asks fill right before flips, and one flip erases ~99 wins. Research only. github.com/Polytutor-AI-Labs…

Round 5, 32/39: Martingale vs Anti-Martingale near 45¢. One adds after adverse moves (reversion bet), the other after favorable ones (continuation). Neither adds edge. Live-ops tell it's dead: adds keep hitting the risk cap before expiry. Research only. github.com/Polytutor-AI-Labs…

Round 5, 31/39: Fibonacci Progressive Sizing. After set triggers, size steps 1, 1, 2, 3, 5, slower than doubling, capped and reset by rule. It picks no side, adds no edge. Live-ops tell it's dead: size sits near the cap more than it resets. Research only. github.com/Polytutor-AI-Labs…

Round 5, 30/39: UDDM. A fixed clock pattern (UP, UP, DOWN, DOWN) picks the side at open, no price signal; size resets after a win, multiplies after a loss up to a cap. Live-ops tell it's dead: a ~50% hit rate while losses stack at the cap. Research only. github.com/Polytutor-AI-Labs…

Round 5, 29/39: LP99M. AutoETS on completed 5-min closes picks UP or DOWN, then fires once at T+1s at the 99¢ entry with martingale sizing. Live-ops tell it's dead: one miss erases ~99 wins, and the martingale makes that miss bigger. Research only. github.com/Polytutor-AI-Labs…

Round 5, 28/39: B5S. Maps last-5-second features (EMA, RSI, book imbalance, CVD, YES drift) to a p_up, compares it with the YES mid, rests a Kelly-sized limit near T−10s. Live-ops tell it's dead: limits fill mostly when the model is wrong. Research only. github.com/Polytutor-AI-Labs…

Round 5, 27/39: CLXM. Projects Binance velocity across an assumed Chainlink lag, compares it with the price-to-beat, one Kelly-sized buy in the last ~2s. Live-ops tell it's dead: high-confidence signals stop beating low-confidence ones. Research only. github.com/Polytutor-AI-Labs…

Round 5, 26/39: CLPTB. In the last seconds, buy the side of the price-to-beat Chainlink is on, allow one opposite re-entry if it crosses, hold to settlement. Live-ops tell it's dead: re-entries keep landing just before price whips back. Research only. github.com/Polytutor-AI-Labs…