Also... add your trade costs; mine are around $2.75. That makes a difference to a backtest that trades 1,000 times or more, though much less than the essential slippage of course.

Also... buy and read Kevin Davey's book "Building Winning Algorithmic Trading Systems". It covers almost all of the biggest pitfalls you're coming across, including the ones explained here.

Testing on blind periods (either manually or using your platform's 'walkforward' mode) is also essential.

Pick a random spot on your equity curve. Imagine you had just started trading at that point. How long before you made money? Now do that two or three more times. Could you have happily traded through the ensuing 3, 6, 9, 12 months?

Did you consider running this in SIM for a while, to see if your results match?

Your average profit per trade without slippage is $31. With a couple of ticks of slips (-$25) this leaves $6 per trade.

Your largest loss is over $7,000 and will increase as you add slippage, and no fills for some limit orders. (Try using 10% of limit entry winners not filled for backtesting. Also a tick of slippage for each leg of a market order.)

The biggest hurdle will be to enter on a limit rather than market. Limits don't always get filled and market entries create slippage between the bid and ask.

Keep us updated as you generate a report with slippage. Best wishes, your hard work is showing.

Well, my 2cent.
I personally would not trade a intraday strategy
that has a 1 year ‚come back‘ period. It only makes
your broker rich
I tell you my KPIs to put a future intraday strategy online.
(Real Money on Micro ES mini for 3months then ES Mini)
It may not apply to others. The out of sample test
must show:
- min 50% winners or better. Less than 40% is a NoNo.
Your mind has to deal with a lot of losers then which will
feel right. especially in the beginning.
- no more than 6 consecutive losers
- max drawdown less than 11%
- sharpe better than 0.8
- sortino better than 2
- avg winner for sure way better than 0.1%
- recover in less than 3 months

You can expect In real money it will perform worse due to changing market conditions.
If it’s automated the strategy will suffer from IT failures, missed trades that were
Pnl positive. Etc.

This sounds for sure hard to achieve but it prevented me
from overall losing strategies.

So start very small, but start

Many good advice here. Hope you follow and re test your strategy.

Your new result including 1 slippage shows 95k profit over 14 years of trading. That is 7300 per year. Your max draw down is at almost 18k. If you double your max dd to trade, 36k commitment will yield you 7300 per year. That is 20% roi.(in real trading, it will be less) and per year, trading 438 round trips, at $8 per, it will cost you 3507.

So. If you take this strategy to real trading now, more chances of it producing unattractive results.

I am not an expert. Trying to make my system work in real trading arena myself but my suggestion would be this.

- limit your losses. Make them a set risk. 7k largest loss is simply not tradable.

- reduce number of rading.

-Research and analyze different conditions. Such as high volatility vs low. Longer term ma falling vs raising, above or below vwap, price above or below certain pivot, yesterday closing ir open etc. Test as many as you can. This would be beneficial to limit your number of trades as well as finding out where your strategy performs best or worse. And test long and short separately. Treat them as a one strategy. This is to understand your system's characteristics behavior.

- Understand how system performs in three major daily conditions. Bull, bear and side ways. This is to find where your system shines at most.

Doing mentioned above made me understand my system much better and made improvements in my system trading without adding layers and layers of filters to fit past data. Hope this helps a little and good luck.

There is another book that I found as helpful (if not more so) as Kevin Davey’s “Building winning Algorithmic Trading Systems”. This is “Trading Systems - a new approach to system development and portfolio optimization” by Urban Jaekle and Emilio Tomasini.

Both books also deal with risk of ruin and how to use Monte Carlo simulation to assess the likelihood of this occurring. Monte Carlo is part of NinaTrader so this is relatively simple to perform now.

Both were incredibly helpful to me in understanding how to build automated systems and the steps required to make a system reliably profitable.