Backtesting irregularities

The sport of kings.
sa7med
Posts: 800
Joined: Thu May 18, 2017 8:01 am

ruthlessimon wrote:
Wed Jul 04, 2018 7:40 pm
r.e. my previous post

So.. I plugged in a basic "trend" strategy, which would provide a back/lay signal, then trade it, exit @ post.

Windsor came back as +42ticks on the real data. Now I've only done a couple of sorts, but the +42 sticks out as something that might be significant. (assuming I've set this experiment up right!)

1. REAL: +42ticks
2. RAND1: -15ticks
3. RAND2: +15ticks
4. RAND3: -9ticks

Now I struggle with the "understanding why" bit. Is this inherent to Windsor (i.e. Windsor just has good trending drifts)? Or is this being caused by an extraneous variable? Did those drifts have high volume? Was MMWAP much higher than my entry (therefore it wasn't a trend)? Etc

Image
Not sure I understand this approach. Could you explain further? How do you "shuffle" the data? What are you shuffling? I'd like to give it a shot
User avatar
ruthlessimon
Posts: 2094
Joined: Wed Mar 23, 2016 3:54 pm

ShaunWhite wrote:
Wed Jul 04, 2018 9:21 pm
I'd be suprised if regional variations and demographics weren't a factor at some level. There may even be a correlation with net disposable income, but the data doesn't go back far enough for people to have had any!
Wouldn't it be opposite? The less affluent places (i.e. North :lol: ) are unformed & like to have a punt. Carlisle, Newcastle, Hexham. But it's also a bit of a paradox, because we could use the uninformed argument to explain predictable reversals
User avatar
ruthlessimon
Posts: 2094
Joined: Wed Mar 23, 2016 3:54 pm

sa7med wrote:
Wed Jul 04, 2018 9:58 pm
Not sure I understand this approach. Could you explain further? How do you "shuffle" the data? What are you shuffling? I'd like to give it a shot
Firstly I began with a basic trend strategy. The strategy takes the price of the fav at x-time (i.e. 03:55 to post). If the price drifts x-ticks we lay the fav, if the price steams x-ticks we back the fav (i.e. 6ticks). The exit always takes place at 00:00:00. & I applied that strategy to a recent 3-mth dataset (Apl, May, Jun).

I then summed the resulting ticks won/lost of the strategy, for each course.

If we take the "REAL Ling"

Steam: 37 - this means there was a profit of 37 ticks backing the fav if it steamed x-ticks from time x @ Lingfield)
Drift: -24 - this means there was a loss of 24 ticks laying the fav if it drifted x-ticks from time x @ Lingfield)

What I then did was disconnect the price data from the course. (i.e. took a Kempton market, & swapped the price data with the price data of a Cheltenham market). This deliberately creates bogus data, which shouldn't be close to the original data (unless the original result was random). We can't make money from anything that's inherently random.

Windsor had an unusually high profitability when it came to laying weak favs - considering its count (12 markets). As shown also via the random tests (Rand1, Rand2, Rand3). They can't get near 42 ticks - suggesting non-random activity

Peter is the true guru when it comes to this stuff. He'd probably look upon this type of analysis as unnecessary, time-consuming & potentially straight up wrong! Hence why I'm hoping he's gonna pop in his thoughts to the thread - & have a more efficient way to solving backtesting issues :)

The name for this type of analysis I believe is "target shuffling" (https://www.elderresearch.com/company/target-shuffling)
Korattt
Posts: 2405
Joined: Mon Dec 21, 2015 6:46 pm

with the greatest of respect people I really do think chasing this line of thinking is making things more complicated than they actually are?
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

ruthlessimon wrote:
Wed Jul 04, 2018 9:47 pm
ShaunWhite wrote:
Wed Jul 04, 2018 9:21 pm
momentum is more likely to occur where the proportion of affluent but largely ill-informed attendees is higher.
Which reminded me of this video :) : https://www.youtube.com/watch?v=e6R4HCwP6nc
well well well, it took Peter 16 years to figure that one out, maybe I do stand a chance afterall ;) ;) ;)
ruthlessimon wrote:
Wed Jul 04, 2018 9:47 pm
.. buuutttt how do we combat this big bad bias (https://en.wikipedia.org/wiki/Look-elsewhere_effect) - i.e. confirmation bias
Do you need to? There appears to a correlation that isn't just random, it's explainable. Effect of variable + Logical explaination = $$$ ?
ruthlessimon wrote:
Wed Jul 04, 2018 9:47 pm
Affluent ill-informed, yeah Surrey would be a good candidate, but there sits Lingfield - which has a gigantic extraneous variable attached. How would we weight those two aspects into predicting momentum?
What's the gigantic extraneious variable with Lingfield? I wouldn't put Lingfield in the same league as Kempton, Sandown & Windsor. It's miles out of town in the sticks and frankly a bit crap. Nobody in town knocks off early to go to Lingfield. Am I right or wrong, is Lingfield like the others? I'm only going on a gut feeling but I'd guess it's an outlier.
ruthlessimon wrote:
Wed Jul 04, 2018 10:01 pm
ShaunWhite wrote:
Wed Jul 04, 2018 9:21 pm
I'd be suprised if regional variations and demographics weren't a factor at some level. There may even be a correlation with net disposable income.
Wouldn't it be opposite? The less affluent places (i.e. North :lol: ) are unformed & like to have a punt.
When you only have a tenner to punt with you make sure you're informed and your tenner doesn't even make a ripple. When you're chucking £50's around like confetti with a nose full of charlie and a bladder full of Bollinger it's a totally different story. Me & Mrs Me used to go, get chatting to the right people, have a bit of a flirt and never have to buy a drink... and we'd go home wrecked carrying half a bottle each and the £20 we'd started with. I've been given (not offered) more drugs there than anywhere I've ever been.... A small bet to them would be £200 (mid/late 90's), one of them might have some clue and everyone would just go on the same or on the one with the funny name. Proper bets say 1 or 2 grand. I remember taking 5 grand to the ring for one guy who couldn't even get off his stool he was so ****-faced. The rich really do know how to do debauchery.

Big money used to change hands at Wimbledon, Catford , Hackney & Walthamstow too. Posh money having fun slumming it with the oiks, and even bigger money being laundered.

This might all sound wildly off topic but it all goes in the mix as Peter pointed out in that vid.

hey how come even my best stories still feel 2nd rate compared to Mr I-dropped-my-sunglasses-in-justin-timberlake's-pool Webb 8-) :o :?
(I'm saving my 'i went to the cinema with mick jagger and david bowie' story, it's my ace up the sleeve ;) )
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

Korattt wrote:
Thu Jul 05, 2018 1:20 am
with the greatest of respect people I really do think chasing this line of thinking is making things more complicated than they actually are?
To some extent I think you're right. But the markets way out on the edge of normal, like wolvs on a wet winter wednesday and the metropolitan courses on a beautiful summer's evening definately behave differently to each other. But am I going to be looking at whether regions have a high density of ABC1s or council estates? Probably not.

Forewarned is forearmed as they say..... i might get that printed on a mug.

...and I might get Simon this one .... ;)
You do not have the required permissions to view the files attached to this post.
Korattt
Posts: 2405
Joined: Mon Dec 21, 2015 6:46 pm

there are some markets I feel l have an edge over others, I stick with them BUT I do monitor the ones I can’t read, in time with experience I’ll gain the confidence to get involved with these, I don’t think you can isolate them by course though,

Yesterday the first half of the day wasn’t really to my liking but after then I pretty much smashed it, if I was confident with reading the first half of the day then it would’ve been quite an astonishing day, that’s where I think as a newbie I used to go wrong, trying to trade every market at every course the same, it’s just not possible, every course will give you a quality opportunity, you just need to know when to pick up on it, but it’s only a quality opportunity if you can read it.
User avatar
ruthlessimon
Posts: 2094
Joined: Wed Mar 23, 2016 3:54 pm

Korattt wrote:
Thu Jul 05, 2018 1:20 am
with the greatest of respect people I really do think chasing this line of thinking is making things more complicated than they actually are?
Absolutely, but there needs to be a balance. Too simple, & the odds of other market participants finding the strategy is high. However, a strategy that is discretionary & incorporates many aspects would be incredibly difficult to replicate.

I personally think most things people use to trade, simply doesn't work. That's why they have psychological issues etc - hence why I try & quantify (roughly), what I trade with (i.e. the above posts). Do I really need this indicator? Why is this indicator so vital?

Talking of complicated, one day you should walk us through just what's going on with your charting Korattt (that's my kinda complex nightmare!) :)
User avatar
ruthlessimon
Posts: 2094
Joined: Wed Mar 23, 2016 3:54 pm

ShaunWhite wrote:
Thu Jul 05, 2018 3:55 am
What's the gigantic extraneious variable with Lingfield? I wouldn't put Lingfield in the same league as Kempton, Sandown & Windsor. It's miles out of town in the sticks and frankly a bit crap. Nobody in town knocks off early to go to Lingfield. Am I right or wrong, is Lingfield like the others? I'm only going on a gut feeling but I'd guess it's an outlier.
The extraneous variable I had in mind was all-weather

I think there needs to be an entire series on "trading relativity". i.e. defining between "trend" & "reversal". My "trend" might be your "reversal" (& vice versa) - simply depends on where we take the start position. With Lingfield it's the exact same. If I divide affluence by county, but yourself "close vicinity" - we could come to opposite conclusions - yet both with valid reasons to argue each case.
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

I don't see the random drawing of lines on a map ie county, as being relevant. Some counties are quite different from one end to the other. The question is why (some of) Surrey, why (some of) Berkshire. The answer might be the abc1s, but Croydon is surrey too and it's broke. So proximity to money seems to trump County.

I'm not sure either how of if all-weather affects it either, apart from the fact that grass under feet is perceived as 'proper racing' for those who like to be seen at the right places. A day at an all weather course feels more like a night at the dogs than it does a day at Ascot.

I agree about defining terms but viewpoint is what makes it so hard. Defining a trend or reversal is like describing an Escher drawing to a blind man.
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

I just picked up this Si from Halliday's post about Aussie racing
Halliday wrote:
Tue Jul 03, 2018 10:32 am
Just added link with details of metropolitan tracks,( better liquidity )
Korattt
Posts: 2405
Joined: Mon Dec 21, 2015 6:46 pm

ruthlessimon wrote:
Thu Jul 05, 2018 8:48 pm
Korattt wrote:
Thu Jul 05, 2018 1:20 am
with the greatest of respect people..
Do I really need this indicator? Why is this indicator so vital?

Talking of complicated, one day you should walk us through just what's going on with your charting Korattt (that's my kinda complex nightmare!) :)
that’s where I was trying to come from last week, to spot a move all the indicators Bet Angel offer that are necessary I really think can actually be built into one indicator, I’m sure it’s possible, the problem then however is that it then compromises an edge for those who are able to read a market & it would undo several years of learning, to have 3 or 4 years under my belt suffering, sorry learning & then for others to be spoon fed an edge I would find most, shall we say.. irritating, so I can understand why an all in one built in indicator isn’t plausible.. or is it?
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

Korattt, I've spoken with people who have put every indicator from betfair and many others into machine learning software and I can tell you that a magic bullet indicator really doesn't exist. Even ai is a very poor substitute for experience.

If it was possible, why would anyone trade manually? We'd all just rent some time on Watson, get our algorithm, and go sit on a beach. Watson would certainly make my spreadsheets look like a flint axe.

We all know someone with the knowledge, the experience, the models and the substantial R&D budget. As far as I know he doesn't use this approach.
Korattt
Posts: 2405
Joined: Mon Dec 21, 2015 6:46 pm

Shaun,

what I find interesting is that there is another professional trader, (and we all know who), someone who I have respect for who runs a course and really only focuses on one edge, one method, one strategy, he does deviate from it from time to time but he likes one particular method, I know of people that have attended his course yet still are unable to execute his method.

What I’m getting at is this, what if everyone on here shared the edge I have learned, and I’m not talking about the one I’ve learned from BA Training day, would it escalate moves exponentially or would it nullify them?
Last edited by Korattt on Sun Jul 08, 2018 5:00 pm, edited 2 times in total.
User avatar
to75ne
Posts: 2413
Joined: Wed Apr 22, 2009 5:37 pm

Korattt wrote:
Sun Jul 08, 2018 4:40 pm

What I’m getting at is this, what if everyone on here shared the edge I have learned, and I’m not talking about the one I’ve learned from BA Training day, would it escalate moves exponentially or would it nullify them?
assuming there was enough of you all more or less entering at a similar point using the same rationale and all exiting at a similar point using the same rationale, i reckon you would in effect be cutting your own throats because you would probably be providing a good oppourtinuity for anyone who can reasonably read the market, to see an exploitable anomaly.
Post Reply

Return to “Trading Horse racing”