Month six. I had a backtest I loved, a bot that ran, and an account that was down 11%. I was sure something was broken. Nothing was broken. I was just six months into a process that takes about three years.
Nobody tells you that part. The videos show a clean equity curve and a promise. They skip the calendar.
Year one is plumbing
Your first year, you won't be trading a strategy. You'll be fixing infrastructure. API keys that expire. A server that restarts at the wrong moment. Order sizes that round the wrong way on one exchange and not on another.
I lost more money to bugs that first year than to bad ideas. A position that never closed because a flag didn't reset. A stop that was calculated on the wrong candle. These aren't strategy problems, but they cost real money, and they're the tuition.
If you judge year one by profit, you'll quit. Judge it by how many failure modes you've seen and fixed.
Year two is humility
Year two is when the strategy itself gets tested, and this is where most of my early beliefs died. A parameter I'd tuned beautifully turned out to be fitted to one stretch of 2021. A filter that looked smart just removed my best trades.
You also hit your first real drawdown. Not the one in the backtest, the one in your account, with your money in it. It feels different. Even with a rule telling you to hold, your hand wants to touch something.
Here's the honest part. Most people who go through year two without a written process change the system mid-drawdown, and that's the end of any chance of knowing whether it works. You can't evaluate a strategy you keep rewriting.
Year three is boring
Somewhere in year three, the work gets quiet. You check the bot once a day, then once every few days. The edge doesn't feel exciting anymore. It feels like a utility.
That's the point where the numbers start to mean something, because you finally have enough live trades to compare against the backtest. Not a handful. Hundreds. A 30% gap between live and tested results tells you very little at 40 trades and quite a lot at 400.
I think this is where the edge, if there is one, actually shows up. Not as a spike. As a process that survived long enough to be measured.
Why people quit at month six
The problem isn't that systematic trading is hard. It's that the expectation is wrong. If you plan on six months, then a flat or negative six months looks like failure. If you plan on three years, the same six months looks like year one doing what year one does.
The sunk time is real, so is the cost, and I won't pretend otherwise. Three years of imperfect results, with real capital at risk, is a serious commitment. Some people shouldn't do it, and that's a fine answer.
What to do with this
If you're starting, set a calendar for the first twelve months and decide in advance what counts as progress. Bugs found and fixed. Failure modes logged. Rules written down. Size kept small enough that a mistake is a lesson, not a funeral.
Then ask yourself one question at month six: did I learn something I couldn't have learned without risking money? If yes, you're on schedule.
At v33 Systematic we publish the full backtest and the assumptions behind it at v33systematic.com, partly so that nobody has to spend year one rediscovering the same plumbing problems we did.