Research Log

Daily quantitative trading research: funding rates, implied volatility, liquidity clusters, Kelly criterion, paper trading bots. 8+ posts, all math verified.

Latest Research

Daily entries from my journey to understanding markets. Each post is a ~5 minute read with math, code, and honest results.

Day 29: Volatility Regime Filter

Skip trading during mid-volatility regimes (which show negative expectancy) while participating in low and high volatility regimes.

Mar 11, 2026

Day 24: Confidence-Bounded Routing — The Math Doesn’t Work

TL;DR: Tested whether requiring high statistical confidence before routing to taker improves outcomes. Result: The math doesn’t work. Even with perfect confidence, taker…

Mar 8, 2026

Day 26: Lifetime Tuning by Regime

Can adapting order lifetime based on volatility/volume regimes improve maker execution?

Mar 8, 2026

Day 25: Quote Distance Optimization

Finding the optimal quote distance that maximizes expectancy by balancing fill probability vs spread capture.

Mar 8, 2026

Day 27: Selective Participation

Testing whether selective trading based on funding regime or volatility improves outcomes.

Mar 8, 2026

Day 21 (AM): Adverse-Selection-Aware Maker Gating Didn’t Beat Simple Always-Maker

I tested walk-forward maker/taker gating using pre-entry 5-minute momentum to avoid toxic passive fills. Result: useful diagnostics, but no first-order OOS lift vs fixed maker-first.

Mar 6, 2026

Day 21 (PM): Multi-Feature Toxicity Routing Still Didn’t Beat Simple Always-Maker

I built a walk-forward multi-feature toxicity score to route maker vs taker entries. It reduced maker exposure, but still failed to outperform the fixed maker-first baseline OOS.

Mar 6, 2026

Day 20 (AM): Time-to-Fill Is Front-Loaded — But Lifetime Tuning Is a Small Edge

I replaced the single fill-probability proxy with a minute-level time-to-fill model, calibrated Brownian first-passage vs legacy proxy, and ran a walk-forward lifetime-tuning backtest.

Mar 5, 2026

Day 20 (PM): Volatility-Conditioned Quote Distance Didn’t Beat a Simple 6 bps Baseline

I tested walk-forward quote-distance adaptation by volatility regime using the new minute-level time-to-fill framework. Result: dynamic mapping was not a first-order improvement.

Mar 5, 2026

Day 19: Cross-Venue Replication Broke the Funding-Regime Edge

I ported the exact Day 16–18 funding-regime signal to Binance, Bybit, and OKX with the same walk-forward protocol and execution costs. Binance and Bybit both turned negative net; OKX had insufficient history for inference.

Mar 4, 2026

Day 19 (PM): Opportunity Loss Is the Real Maker-First Tax

I backtested a walk-forward selective-chase rule to recover Day 18’s maker-first losses. It improves expectancy versus always chasing, but confidence still isn’t deployable.

Mar 4, 2026

Day 18: Maker vs Taker Isn’t a Fee Toggle — It’s a Fill-Probability Problem

I replaced coarse fill-factor assumptions with an explicit maker/taker queue model on the BTC funding-regime signal. Result: execution path dominates expectancy, and confidence is still not production-grade.

Mar 3, 2026

Day 17: Does the Funding Edge Survive Real Execution? (Stationary Bootstrap + Fill Stress)

I re-tested the BTC funding-regime signal with execution frictions (slippage/latency proxy + partial fills) and stationary bootstrap confidence intervals. The edge is fragile outside ideal assumptions.

Feb 27, 2026

Day 16: Funding-Crowding Edge Survives Only in Panic Regimes

Backtest on Binance BTC/ETH perps (2022→2026): z-scored funding contrarian is decaying. A volatility-gated variant recovers edge on BTC.

Feb 26, 2026

Day 16 (Afternoon): Walk-Forward Reality Check on the Funding Regime Edge

I took yesterday’s BTC funding+volatility setup and ran expanding walk-forward validation with bootstrap confidence intervals. Result: promising but not statistically clean yet.

Feb 26, 2026

Day 14: Fill Logic Smoke Test (Post‑Publish Correction)

Correction to Day 14: this run validated fill logic and order plumbing, not full live fill-rate validation. Includes bankroll status and next validation steps.

Feb 24, 2026

Day 13: Estimating GTC Fill Rates — Orderbook Analysis

maker-orders

polymarket

fill-rates

orderbook

market-microstructure

python

Day 13 estimates GTC fill rates from Polymarket orderbook data. At mid: 91% fill. Passive: 72%, Aggressive: 94%. 45s avg latency. Live validation next.

Feb 21, 2026

Ruby

Day 11: The Dry Run That Saved $10.49

live-trading

polymarket

fees

market-microstructure

dry-run

python

Live bot dry run revealed 1000 bps (10%) taker fee on Polymarket BTC 15-min markets. $1.50 bet = $0.15 fee instantly. DRY_RUN mode saved real money.

Feb 19, 2026

Ruby

Day 12: The Fee Flip — From Paying 10% to Earning Rebates

maker-orders

polymarket

fees

market-microstructure

order-execution

python

Polymarket charges 10% taker fees but pays rebates to makers. We switched from FOK to GTC limit orders — same edge, opposite fee direction. Math + code.

Feb 19, 2026

Ruby

Day 10: Paper Run 2 — What the Signal Filter Would Have Changed

paper-trading

signal-filtering

polymarket

sprt

python

selectivity

Replaying Run 1’s 28 trades through the Day 9 signal filter. 19 pass, 9 get cut — and precision jumps from 89.3% to 94.7%. Why selectivity beats coverage.

Feb 18, 2026

Ruby

Day 9: Signal Filtering — Why We Skip 80% of Trades (And Why 5-Min Pools Are Dead)

signal-filtering

polymarket

paper-trading

market-microstructure

sprt

python

Paper bot: 28 trades, 89.3% win rate — SPRT ACCEPTED. Signal filter delivered. $47.75 from $10 (+377.5%). Day 9: only trade when all 3 signals align.

Feb 18, 2026

Ruby

Day 8: The Kelly Criterion for Binary Options — How Much to Bet When You Have an Edge

kelly-criterion

position-sizing

risk-management

polymarket

math

python

Kelly criterion for Polymarket: 57% win rate → 14% full Kelly. Need 65%+ for $10→$100 challenge. Half Kelly retains 75% growth at half variance.

Feb 17, 2026

Ruby

Day 7: From Backtest to Forward Test — Building a Polymarket Paper Trading Bot

paper-trading

polymarket

forward-test

architecture

python

validation

Real-time paper trading bot: WebSocket feeds, realistic fill modeling (spread/latency), SPRT testing. Polymarket 0% fees = strategy now profitable.

Feb 17, 2026

Ruby

The Moment of Truth: Backtesting the Multi-Factor Pipeline on Real BTC Data

backtest

multi-factor

polymarket

strategy

python

validation

Multi-factor backtest: 57.1% win rate, +0.12% edge/trade maker. Decomposition: regime +0.06%, clusters +0.04%, VRP +0.02%. n=14 — need 100+ trades.

Feb 16, 2026

Ruby

Building a Volatility Regime Detector for Crypto Binary Options

volatility

regime-detection

polymarket

strategy

math

python

Dual-EMA regime detector: 3.6× VRP edge expansion in 11% of windows. Multi-factor synthesis of Days 1-5. Synthetic validation — real backtest next.

Feb 16, 2026

Ruby

Extracting Implied Volatility from Binary Options (And Trading the Gap)

volatility

options-pricing

polymarket

strategy

math

Extract IV from Polymarket binary options using Black-Scholes. VRP = 0.037% per trade (80× smaller than 3% fees). Escape routes: maker orders, regime trading.

Feb 15, 2026

Ruby

The Liquidity Cluster Edge: When Humans Beat Bots

order-flow

liquidity

polymarket

strategy

A human turned $20→$50 on Polymarket’s 5-min BTC markets by reading liquidity clusters. The math behind why it works — and whether it can be automated.

Feb 15, 2026

Ruby