Why I stopped trading and started tendering with AI

Description and use

The author moved from predictive trading to AI‑driven tendering, scoring opportunities with volatility and adjusting risk tolerance via PPO reinforcement learning.

Problem

Consistently predicting market movements is difficult and fragile; small changes in volatility, regulation, or news can break trading strategies.

Outcome

The transition from prediction‑centric trading to AI‑driven tendering reduced sensitivity to market regime changes, introduced risk‑aware scoring, and leveraged RL to adapt risk tolerance, potentially improving sustainability and profitability.

Workflow

• 1. Compute short and long moving averages from price data.
• 2. Calculate 20‑day volatility as a risk indicator.
• 3. Assign a tender score: reward higher volatility on buy signals, penalize on sell signals.
• 4. Compare score to risk_tolerance to decide BUY, SELL, or HOLD.
• 5. Return signal and score for further filtering.
• 6. Use a PPO RL agent to predict adjustments to risk_tolerance based on market state features.
• 7. Execute trades with position size scaled by tender score when signal is BUY or SELL.

Evidence

I poured countless hours (and frankly, too much money) into building systems that promised to consistently outperform the market.

Source

Instead of predicting the market, I started focusing on responding to it in a more strategic, less frantic way.

Source

I used a Proximal Policy Optimization (PPO) agent, trained on historical market data, to dynamically adjust the risk_tolerance and weighting of the tender score components.

Source