The Wake-Up Call
How we cut max drawdown in half while protecting gains across 10 volatile tickers — and what that taught us about the difference between winning and surviving.
We were crushing it. Our system was up 5.09% with solid realized gains. Then we looked at one number that stopped us cold: our worst-case unrealized loss was nearly as big as everything we'd made that day.
That's the gut-punch moment in algorithmic trading. It meant that if every trade's protective stop triggered right now, we wouldn't just give back our gains—we'd be down for the day. A winning session could become a disaster before lunch ended.
That's when we realized we weren't actually protecting our profits. We were just hoping the market would cooperate.
The Problem We Were Ignoring
Our system used a fixed 4x ATR trailing stop (ATR stands for Average True Range—basically, how much a stock typically swings in a given period). We set it at entry and let it follow price up. But here's the issue: the distance from price to stop never changed, no matter how much money we were making or how the trade was moving.
It's like having a seatbelt that only loosens as you accelerate instead of tightening.
Three Approaches to the Fix
Phase One: Step-Down Trailing Stops
We started at 4x ATR on entry, but once we hit +1.5x ATR in profit, we tightened the stop to 2.5x ATR. Then at +3x ATR profit, we moved it in further to 1.5x ATR. The key: we used the current ATR, not the one from when we entered, so the stop stayed responsive to real market movement.
Phase Two: Partial Exits
We added another layer: close half the position at +1.5x ATR, then move the remainder to breakeven. This way, we're banking profit early and removing the pressure to hold a full position. Sometimes getting half and walking away beats getting greedy and getting nothing.
Phase Three: Stall Tightening
If a trade was sitting nearly flat for 30 minutes—making less than 0.5x ATR in profit—we automatically tightened the stop to 2x ATR. A stalled trade is a trade that needs to prove itself or get out of the way.
What the Numbers Actually Say
Over six months, RIOT went from $1,586 in cumulative P&L to $4,368. RGTI jumped from $3,170 to $5,437. MSTR nearly doubled from $1,641 to $3,167. Win rate climbed from 47% to 70%, and max drawdown got cut by 50-60% across the board.
On longer backtests, SOXL told an even cleaner story: same final profit ($15,620 vs the baseline $15,402), but max drawdown fell from $1,303 to $661. Half the pain for the same destination.
IONQ showed us that tighter stops could actually improve returns when you're sharp about it: $12,737 baseline became $17,507, a 37% boost. Starting with a simulated $10K account, we'd end up in roughly the same place—but with a dramatically smoother ride getting there.
The Metric That Changed Everything
We started tracking what we call "Protected Ratio." It's a simple calculation: your worst-case unrealized loss divided by your unrealized gain. If that ratio is negative (like our original -1.38), it means a sudden reversal could wipe out all your open profits and then some. It's a wake-up call.
After we implemented these three improvements, that ratio went positive. Your worst case became profitable, not catastrophic.
The Real Lesson
Winning the market is one thing. Keeping what you win is another. We learned that the best trading system isn't the one that catches the biggest moves—it's the one that lets you sleep at night knowing your downside is locked down.
The improvements took a few weeks to build and test properly, but they turned a system that felt fragile into one that feels built to last. And after watching ourselves nearly turn a 5% day into a loss, that peace of mind was worth every hour of work.