Recently debugging a quantitative strategy, the thought process has been quite complex, and I want to organize the entire process.

**Growth Trajectory of Three Iterations**

The first version of the strategy was tested for 1 day, and it directly made a profit of 20u, which was quite exciting at the time. After the second version went live and ran for 6 days, its performance was even more impressive, accumulating over 80u in gains. It seemed like it was about to take off, but the third version started with a crash—during the 5-day testing period, it experienced continuous declines, with the deepest drawdown reaching an unrealized loss of even -100u. However, what's interesting about this version is that it has a clear drawdown recovery capability.

**Current Account Status**

The day before yesterday, the realized profit and loss was still at -18u, but yesterday it directly turned around to -5u. More importantly, the unrealized profit doubled from 12u to 25u. If all positions are closed now, the third version could lock in a profit of 20u for this week; if it can hold back to the initial opening price of 1.95 (the average price), it could unlock 100u. Calculating over these approximately 13 days, the current achievable profit is 120u, and in an ideal scenario, it could reach 200u.

**Contradictions in Drawdown Management**

To be honest, the current drawdown data is not ideal. In the past half month, the maximum drawdown reached 30%, which is an estimate based on the extreme continuous declines seen in the third version. Using an initial margin of 300u, with a peak unrealized loss of 100u, the drawdown rate is about 25%. If we include the existing unrealized profit (which inflates the account to 400u), the drawdown pressure becomes slightly more manageable.

I know that a 30% drawdown line is somewhat tight; ideally, it should be controlled within 20%, but that level of difficulty is indeed very high. The current approach is to continue optimizing parameters to see if we can maintain the win rate while reducing individual risk exposure.