The opinions expressed in these forums do not represent those of C2, and any discussion of profit/loss is not indicative of future performance or success. There is a substantial risk of loss in trading. You should therefore carefully consider whether such trading is suitable for you in light of your financial condition. You should read, understand, and consider the Risk Disclosure Statement that is provided by your broker before you consider trading. Most people who trade lose money.
This is why forward testing is more important than backtesting. Backtesting only shows what the strategy CAN do but not necessarily what it WILL do.
I only use backtesting to figure out if the strategy is viable or not. Then I forward test it. You will then understand the weaknesses of the strategy.
Most likely one strategy will NOT work under every situation and as such will have a period of drawdowns. If the strategy drawdown is acceptable then you have a winning system.
Hope this helps.
If I had a dollar for every time someone asked me some version of “why don’t you just use tighter stop losses?” during a drawdown…
I REALLY REALLY dislike stop losses… they screw you 95% of the time, until the one time they save you from losing everything of course…
@ChrisPage, I agree using physical stops can be hit because tools now can allow stops to be hunted.
But what you can do is use programmable stops that allow the order to exit once past a certain level. Also you can use derivatives to hedge.
Using these techniques is what separates the amateurs from the professionals…lol
Hope this helps.
Are you nostalgic too, and remember the good ole days when weak payroll and jobs data would actually be a catalyst for the market to go DOWN instead of UP…
Our AI strategy generation and tuning has access to BILLIONS of data points going back to the beginning of 2020, but it does not have access to any (direct) sentiment data. However, I SWEAR that as I watch it trade, sometimes it’s like it’s using “old school” (i.e. pre Fed manipulation/tightening) fundamentals to make trade decisions.
The market trended flat or down most of the night overnight, with weak jobs and payroll data and the BoJ saying the market is too optimistic about rate cuts, and pretty much all headline data pointing to economic headwinds, and our portfolio went short. Then, pre-market this morning, straight up the market goes on more weak data and the HOPE that we have a weak payroll report tomorrow, and our “hey, there’s bad news on the horizon” short stopped out.
From Reuters after the latest data release this morning: “Reports showing weak private payrolls and job openings this week have reinforced expectations the Federal Reserve’s furious pace of rate hikes is slowing the economy, potentially allowing the central bank to ease up on its monetary policy next year.”
Even if we do figure out a way to ingest scored sentiment data, which we HAVE looked at, the real trick will be our models knowing when good news is ACTUALLY good news, and when good news is actually bad news (and vice-versa)… UGH.
We’re testing new logic that basically ingests a calendar of scheduled releases of federal economic data, and minutes prior to and following those data releases, momentarily modifies the stop loss orders of any currently open positions to hopefully not get “wicked out” of an open position during the initial release-driven kneejerk price movement.
In backtesting, this feature yields higher positive ROI in 38 out of 47 trades/events, sometimes dramatically so, and higher ROI EVERY quarter over the three years we backtested.
Interestingly, and perhaps why our A.I. tuning wasn’t already doing something like this, is that this logic actually slightly lowers one of our heavily weighted key metrics, ROI/MaxDD (return divided by maximum observed drawdown). When we dug into the data, the reason is simple: allowing an open position to have a larger momentary drawdown during that kneejerk wick increases our MaxDD value, albeit literally only for a few seconds, but nonetheless skews our ROI/MaxDD calculation lower.
Hopefully going forward we will see fewer instances of stopping out of an open position during the initial market spasm of scheduled economic data releases.
The market: “I see you just added to your long position… hold my beer…” 
I hate giving back some of our December gains in the last three trading days before Christmas, but there’s only so much any automated trading can do when day after day looks like this, a slow steady climb up, and then off the cliff mid-day like lemmings…
Wednesday:
And today…
After a decent “Santa Claus rally” week, looks like we’ll be ending on the whimper of weak Chicago PMI data rather than a bang. Despite our longs stopping out on this dump, we’re still positive for the week and the month, and will be wrapping up 2023 with a solid return for the year…
We’ve had a great last few days, up around 5% for the week and well ahead of the S&P for the start of the year.
That being said, for two days in a row now we’ve been in a small long position that is doing well until it gets stopped out on a pullback in the final minutes of the trade day (both days at exactly 3:50pm), right before the market reverses back up going into the close.
As soon as possible, we’re going to test some logic around increasing stop losses, at least for long positions, the last X minutes of the trade day, likely including other criteria like market behavior up to that point for the day (i.e. if it’s been trending up all afternoon and is positive for the day, expect a late day pullback and temporarily increase the stop loss of any open longs going into the close, or something like that).
Not sure if that’s a winner long term, but that’s what backtesting is for. It sure would have helped this week.
Let me know if you’ve tried anything like this yourself in day trading, and if so, what you tried and how it went!
Since I tend to mostly share our challenges here and what we’re working on to overcome them, I thought I’d share something positive…
In my real job we usually run anywhere from 4-8 automated strategies in a client portfolio instance (some long only, and some short only), each trading whatever share of the portfolio it’s allowed to, mostly based on past performance. I trade a subset of those same strategies here.
In the best cases when everything is falling into place, in a week we’ll see about three quarters of the running strategies in the green, and the rest in the red. That’s by design: different strategies do well in different market conditions, and the portfolio is designed such that they balance each other out, ideally netting a positive return overall regardless of what the market is doing.
It’s not very common to see ALL running strategies in the green toward the end of a week, which is where we are at the moment, up over 6% and ahead of the S&P by 5% for the week…. we all know that could turn on a dime, but it’s nice to see ALL green on the backend 
.
(returns so far this week for a client account trading around $750K, but all clients are well in the green)
LOL no sooner do I post that than we close a great short trade, adding another 1.5% ROI to our week while the market dumps going into the European close… now up almost 8% for the week, ahead of the S&P by 7%…
This should be mandatory in all trading systems…lol…Kudos that you’ve added this.
But changing stop losses can be tricky using programmable logic but its easy to do manually since daytraders just visually watch for support and resistance levels.
Another logic you should add is using multiple contracts with runners having wider stops.
Hope this is helpful.
For anyone following along, I have created a new thread for 2024 for this strategy, and will posting all future updates here: SP 500 Futures Scalper 2024 - New ATH and 50% off
