称用 Grok Bot 把 1000 美元做到 26 万

Logics @immortalhowwl原推发布于 2026.10.26

在 X 上打开原推

A Chinese quant used Grok Bot to turn $1000 into $263,396.56. He studies applied mathematics and builds algorithms between classes. His account: BoneOhio. At such a young age, he has already made over a quarter of a million simply by knowing the right formulas and knowing how to use Grok Bot. Result: $1000 → $263,396.56 Profile: https://polymarket.com/@BoneOhio?via=immortalhowwl How to build a system like this: Quant Flow Memory reads order flow and matches it against historical price movements. The other modules check risk and execution. 1. Multilevel Flow Flowₜ = Σ wₖ × OFIₖ,ₜ The bot tracks new orders, cancellations and fills across several order book levels. One large buyer can disappear. Pressure across multiple levels gives you more information. 2. Flow Memory Responseₕ = Σ gⱼ × Flowₜ₋ⱼ The model identifies order sequences that preceded price moves. The weights are calibrated on historical data. The order book gets a memory. 3. EV Gap Edge = expected move − trading costs Projected move: 3¢. Entry and exit: 1¢. That leaves a 2¢ edge per share. If costs eat the difference, there is no trade. 4. Kelly Criterion f* = (bp − q) / b Win probability and payout determine the fraction of capital to risk. Fractional Kelly reduces exposure to model error. One bad entry should not wipe out the account. 5. Bayesian Updates P(H|D) ∝ P(D|H) × P(H) New fills. Large orders cancelled. Pressure shifts. The bot updates its probability estimate. If the signal disappears, the order is pulled. 6. KL-Divergence Dₖₗ(P‖Q) = Σ Pᵢ ln(Pᵢ/Qᵢ) Measures the divergence between comparable probability distributions. Flags an anomaly for review rather than calling every difference an arbitrage. 7. Stoikov Model r = s − qγσ²(T − t) Adjusts the reservation price for inventory, volatility and risk. The more inventory already held, the more cautious the next entry. 8. LMSR C(q) = b ln Σ exp(qᵢ/b) An additional model for price simulations. Actual entries still need to be checked against Polymarket’s order book. In practice: → Order flow generates a signal. → Memory estimates the move. → Costs and risk pass the checks. → Only then is an order placed. The math is public. The edge needs to be validated on data. The difference is that most people never build the system. Paste these formulas into Grok Bot and build your own bot. Start with simulated trades. Copy him: https://predict.bananagun.io/?referral=ogfT9LaJ Bookmark this post so you don’t lose it. Soon I’ll break down another bot and the formulas behind its strategy.

这条是引用着另一条写的：x.com/rohonchain/status/2092626965224415557