I think absence of a backtesting feature does limit the profitability of many BA customers.
For example, someone has an automation making £10 a day.
What if you could then test it with 500 different iterations of the parameters and optimise its profit ? Ctrader for Forex does this.
Currently people are spending months testing a couple of iterations maybe improving to £12 a day. When with the right tools it could be a £100 automation overnight.
Then not to mention working automations that start on a downswing of variance and get discarded and the user goes in a cycle of years chasing other automations when they had one that worked from the start.
Feels like we are making life difficult for ourselves. Of course I'm aware of the risks of backfitting. But a tool that optimises our parameters could change peoples side hustle into a profession. And it will educate people very quickly on what strategies simply dont work and save them years of going down a rabbit hole wondering if its variance or a bad strategy
While there are problems creating profitable strategies through backtesting with getting bets matched, it can certainly save years of trial and error with strategies that will never ever work. I posted a thread just yesterday that stopped me spending months blindly backing/laying at BSP on greyhounds because I backtested.
BAMI - Bet Angel Market Import
i would add. I'm primarily looking at backtesting as being a way to improve parameters on mechanically based strategies. However, it goes without saying that form based strategies can also be backtested in terms of optimising bet placement (offer price, entry time, exit time etc, etc). Also, backtesting could be used to test blanket coverage on markets/events to see when to avoid and when to commit - etc, etc
I'm always surprised by people asking for this as one of the biggest errors I see people making is overfitting historical data.
You think you have found something but when you put it into practice is loses money. People spend so much time pouring over data that doesn't deliver much value.
I understand the appeal, but it's flawed in my view. All my best systems have come from testing the market and finding stuff that never appears in historical data.
Of course, we also have this problem that Betfair want to charge us an arm and a leg for the right to use historical data. We just don't have the money for that.
For me I’d use it to accelerate pattern recognition.
I’m not tech savvy enough and struggle with a simple spreadsheet at the best of times. You’re probably right in the respect no edge can be gained entirely though this and due to my lack of ability I have no experience to say differently BUT I’m consistently able to gain a slight edge due to patterns I’ve identified through screen time.
If i was able to replicate that screen time at a time of my choosing rather than when the event was inplay I’d be confident I could spot more opportunities.
Totally get the argument for the cost prohibitive.
If it ever becomes a possibility I certainly think it would add to the product.
I’m not tech savvy enough and struggle with a simple spreadsheet at the best of times. You’re probably right in the respect no edge can be gained entirely though this and due to my lack of ability I have no experience to say differently BUT I’m consistently able to gain a slight edge due to patterns I’ve identified through screen time.
If i was able to replicate that screen time at a time of my choosing rather than when the event was inplay I’d be confident I could spot more opportunities.
Totally get the argument for the cost prohibitive.
If it ever becomes a possibility I certainly think it would add to the product.
Last edited by Beagle on Thu Jul 18, 2024 2:16 pm, edited 3 times in total.