Back testing method/software

[Fugazi]

Funny you should post this. I arrived at a similar strategy purely through trial and error and it is why im wanting to learn to data collect. As unfortunately im not quite profitable the way im doing it. Ive come close through trial and error, but even if i manage profitably it wont be anywhere near as optimal as using actual data. Without the data of when exactly the price movements happen im having to green up instead which sucks for EV and often is frequently mis timed leading to a loss.

yes -you really NEED to be partying on top of the stddev as they occur in this type of test (and tbh, there's a few other confluence items to consider, which again requires several additional ways to chop up the current and near term data elements). I found that even tho i was easily(ish!) discovering the acceleration/deceleration points on selections, i was only ever at best making 40-50% correct entries. the additional confluences managed to notch that up a bit, tho at the same time, obviously reduced the number of candidates per event.

I discovered it through ladder movements - one shoots up, another must go down. But I'm missing a few x factors to get me into the green.

[Euler]

Backtesting / backfitting is very different from modelling.

backfitting = delusional approach to believing that historic data can predict the future
backtesting = informed approach to understanding market dynamics and appropriately using data models to ASSIST in execution

I think people often conflate the two. I collect data and lots of it, just to test assumptions or to understand the market better.

But I need to model the data to predict the future and deploy my strategy into it.

if we want to be technically correct, what we actually do is the following:

1. collect and prepare data that forms attributes within a data model
2. create a strategy that is aligned to the attributes within the model
3. simulate the profile of the strategy using a blend of 80% exposed data and 20% blind data
4. run the model against the 80% exposed data (parameter adjustment permitted at this stage)
5. rerun against the blind 20% to ascertain that we have not data/parameter fitted our assumptions
6. randomly re assign a further split on the data and re run as required
7. forward test using either a test account or with small stakes
8. scale into the live market - risk dictated by metrics obtained from backtesting (drawdown etc)

I think we are all saying the same thing, except, I firmly am of the belief that there is a widespread misconception that backtesting equates to looking at form data and running a spreadsheet against that and then creating a strategy. In my world, it's a purely mechanical process which starts with an exploration of the data and then moves onto looking at ways to exploit any observations (an example might be one where the data shows that 95% of the time, there is a huge swell in volume/prices around the -120 second mark, followed by many equally spaced standard price deviations from the previous mean on key runners - etc). Using this *data first* approach, every care is taken to mitigate data fitting and only focus on outlier behaviours etc.

I'm fairly risk averse (but have courage in my convictions). So for me, based on my skillset, the above gives me confidence when i reach points #7 and 8 above.

[edit] - current example which has reached #7 above:

Screenshot 2024-04-03 162619.png

Y, we are on the same page generally. I just see, generally, everybody thinking that looking at historic data is going to somehow predict the future, when it doesn't.

[Euler]

The only thing I would add is that I often come up with ideas and concepts and test them directly in the market.

I know where to look for inefficiencies, so I dig around in those areas and test and probe my assumptions directly in the market.

Those are where all the best opportunities I've ever found have come from. Things that are not apparent or can be seen in data.

[jimibt]

Y, we are on the same page generally. I just see, generally, everybody thinking that looking at historic data is going to somehow predict the future, when it doesn't.

i'm definitely purely observing market behaviour and repeating patterns, definitely not stuff like how many days since last run, what color of top the jockey is wearing etc...

[Euler]

Y, we are on the same page generally. I just see, generally, everybody thinking that looking at historic data is going to somehow predict the future, when it doesn't.

i'm definitely purely observing market behaviour and repeating patterns, definitely not stuff like how many days since last run, what color of top the jockey is wearing etc...

Well, I'm glad you haven't found my jockey colour edge just yet

[Fugazi]

Y, we are on the same page generally. I just see, generally, everybody thinking that looking at historic data is going to somehow predict the future, when it doesn't.

i'm definitely purely observing market behaviour and repeating patterns, definitely not stuff like how many days since last run, what color of top the jockey is wearing etc...

Well, I'm glad you haven't found my jockey colour edge just yet

Interestingly, the colour of the dog in greyhound racing genuinely is significant. I nearly fell off my chair in surprise at it.

[MemphisFlash]

A dog with only 3 legs is also vastly important, for a start it is 25% lighter,
Lighter means Faster, more stream line, etc

[andy28]

A dog with only 3 legs is also vastly important, for a start it is 25% lighter,
Lighter means Faster, more stream line, etc

That must be why they geld horses, they are 2 stone lighter, and 2 stone is a lot in horse racing

[conduirez]

A dog with only 3 legs is also vastly important, for a start it is 25% lighter,
Lighter means Faster, more stream line, etc

That must be why they geld horses, they are 2 stone lighter, and 2 stone is a lot in horse racing

What a load of bollocks! Well lack of them anyway!

[ANGELS15]

I recall one of my own experiences of this from 20 years ago.

I had subscribed to a web database called 'Flatstats' which I believe is still going. It contains historical data on races going back many years.

I had observed that it appeared that top weight flat horses in 6 runner or less handicaps performed quite well. Or so it seemed. I had backed a couple of these which had won. I decided to run the criteria through Flatstats which at that time could go back 10 years. Lo and behold these top weights did show a profit.

Now an interesting thing is that in those days there weren't that many 6 runner middle dstance handicaps going back over the 10 years. Perhaps a couple of dozen a year.

I decided to use it as a betting strategy. The first bet won but all the others lost.

I realised with time and experience that those previous horses had won simply because they were good horses and were entitled to win and not just because they were top weight.

As Euler said it is important to know why. It is also important to put things into context. Classic example Jockey A shows a profit for his rides at Catterick over the last 5 years. When you examine his rides he's the stable's 2nd jockey and gets the lesser rides. One day he was riding a 100/1 outsider in a 6 runner field. 3 of the runners had pulled up and he was 30 lengths behind the favourite and second favourite who looked to have the race between them. The 2nd favourite jumped across the favourite unseating it's rider and causing the favourite to crash through the wings of the hurdle. Jockey A was left to come home alone on the 100/1 chance.

[conduirez]

The link below is a program made by Derren Brown, who exposes many scams, but this one is about horse racing and shows just how easily people can be led, by being given false information to make them believe he can actually guarantee the winner of many horse races.

https://www.bing.com/videos/riverview/r ... ajaxhist=0

If anyone does not know who Derren Brown is, it may seem at first he is a trickster, but he is a television presenter who shows people how easy it is to be scammed.

[andy28]

I saw this first hand with a horse I owned, My horse was trained by Te Akau and was to be ridden by Opie Bosson and in NZL if you see that combo many just pile it on no matter what, before the odds came out the trainer and us owners were discussing his future, the trainer was convinced he was a 1 win horse, where and when he would win he couldn't say but he said that they would give him 3 runs to see what he does. He had been to the trials and performed poorly.

I thought given the trainer/jock combo he would be at $20, he opened at $3.2, not only that the next day he was at $3.00, day of the race (which I didn't attend) and he gets the drift and goes out to $6.00. I wish I could have layed him on BF as he got out to $8.00. Needless to say he run 12th of 15 I think, jockey said he hated the firm ground and it felt like he was running on marbles

But people blindly backed him just because who trained him and was rode him. The trainer was right he won his 3rd race at 60.00 ( which I got at BF) in the slop, won like an odds on fav, but developed back issues and he never raced again. Lost a small fortune on him, but that's racing. Bosson wouldn't take the ride, so stable apprentice got the ride

The TAB have said they have a Te Akau/Bosson algorithm and the horses (especially first starters) will open up very short as they attract money from the Tangerine Army as they are called

[decomez6]

I know where to look for inefficiencies, Things that are not apparent or can be seen in data.

is data without information just noise ?

with minimum preparation between opening and closing trades ,cold traders will still need to keep a record of their trading activities {data} , in order to make sense of the market noise .

Data refining with market experience , will finally result to a personalised indicator.

[firlandsfarm]

If you want to test whether your bot works, then do the test for pennies or in practice mode, however do not use the bot fully until you test it on the live market for pennies, not in practice mode.

Only use Practice Mode to test the rules are performing as expected and certainly do not use Practice Mode to test the profitability of a bot. I have seen profitable practice bots turn to loss making as soon as they go live!

[andy28]

I have found this thread very informative, I split my data like I have always thought but with the use of ChatGpt it showed what I should be looking for and how to achieve it. All I can say is Wow, it transformed my data, while it was profitable it will almost doubled it going forward. I have nearly 750,000 lines of Greyhound result to go thru now, I looked at them and saw nothing so now fingers crossed I will get some thing in there.
The penny might be dropping