When Do You Give Up On A System?
 

As per the title, how do you know if/when a system has had its day? The answer, I suppose, is pretty obvious - when you run out of money! But really, is there a certain point when you no longer have confidence in a system?

I have been betting a system since 1 October, 2013. I won't go into details suffice to say it is a laying system, one horse per qualifying race. There have been 2616 bets up to 4th Jan (I haven't updated my overall records since then but I'd imagine it would be over 3000 bets now).

I started off laying at $30 liability using a ratchet staking system and have had it up to $60 liability. Due to recent losses I have re calibrated the bet amount at the start of January at $40 liability and am currently at $48 liability. It should really be at $38 liability if I wasn't using ratchet stakes.

The bank has gone from $0 to it's highest point of $1390 on 21 Dec. Since then the bank has dropped down to $800 with a $95 loss today. Another $100 loss will see the bank down 50% from it's high.

Should I call it quits there (at 50% down from the high)? Should I do it now?

I remember someone saying on here (I'm sorry I can't remember who) that he never bets over the Christmas/New Year period. I wonder whether that has got anything to do with it and also when that person starts to get back into it?

I have attached a graph.

Well, what would you do if you were in my shoes? (This is the only system I have got so I can't even keep it up and offset the losses against any other system).

I would :

1. go to when the system started losing and do a chi test from that point
2. compare the result to the overall chi test results (all selections)
3. Find your median odds in each range
4. Find your average odds in each range
5. Find your strike rate in each period
6. What's your amounts of bets per week in each period

Compare the results. What has happened ? Do you still have an edge and its just been an unlucky run ? Do the tests show you that you no longer have an edge.

What was your median and average dividend range before when it was winning compared to when it was losing ?

Has the amount of bets changed per week?

Has the strike rate changed ?

I just looked and saw it was only 113 selections. There could be your problem. Its not enough data to any analysis on really to decide if you have an edge and whether it has gone. you really need 300+ for testing, 300+ for paper testing and then evaluate a system after 300+ of real life testing. 113 means you should still be in the testing phase or paper testing.


If you can do the above tests and post your results and I might be able to help you.

I have no idea how to do those things however 113 is the days (112 actually) and not the selections. There are over 3000 selections.

Racing happens 7 days a week 363 days a year. It however is a very seasonal sport. The beauty of time is that you can test and adjust. Back some time ago I subscribed to a tipping service. He never charged for part of September and all of November. He gave the tips on the proviso that they were at one's own risk.

The reason being that in 25 yrs of tipping these weeks never produced a profit. and a lot of the time pulled the annual result down.

Ask yourself what has changed.

Hi Echo,

What sort of testing did you do prior to the start of betting?

Michal

A ratchet system will magnify profits, but also losses.
I would run a test using flat liability and compare the profit/loss over sections of time. It's most likely that the peak in the graph was simply a better than average time and just after was a correction of sorts.
However, the Betfair markets are fluid and edges are constantly eroding, so that the market is becoming almost 100% accurate pre commission. We've lost the big traders which used to push prices, and so all that's left (by volume) are opinions.
It depends how much previous data you had, and how much current data you have as to if or when you bail out. If you're worried, sit it out recording results still if you can (as you are still in profit) and if it returns to the previous high, you'll lose that profit and possibly make a bigger loss getting back on, but your confidence will be restored.
Perhaps wait for the next downturn, but if you plan staying with it long term, it will all even out anyway.

So when do you give up on a system ?

The answer is - there is no answer (IMHO)

As i said somewhere else i have really good and deep horse racing form data that goes back in different periods over 20 years back to 1993.

originally it was part of a quite sophisticated database that was called TK-TOOLS.

In that time i developed systems that would win for 2.....maybe 3 years and involve analysis of '000's of races and 10's of '000s of horses and would extract enough starters to win handsomely ( say 15% POT) - I would be saying - eureka - this is it - i have found it.....but then they would start to get wobbly and then plunge into oblivion and never recover - for no apparent reason.


Others would be rocky from the start and come good for a while and then drop back.

I couldn't work out what changed and couldn't find any filters etc to trim the dive.

In the real world now - I am going through a rough patch right now with my selections. So much so that I have cut my bet size to preserve my bank and be there when the improvement comes (if ever).

What I have done also is examine my results at a deeper level.

In my data I obviously record the results and the dividends etc. – but I go 1 step further – I record the beaten margins for all starters.

What I have found is that whilst my win POT is negative my about 20% ( down from about +10%) - my selections finishing within 2 lengths of the winner are only down slightly – at this point I am hoping that the negative POT is an anomaly and they are thereabouts – but just not winning and that it will improve later on.

Hope springs eternal as they say.

So that might be one way - analyse your actual results and see how close they are.
Maybe you are just getting beat by bad luck - which could eventually turn around.

Since your system is a lay system TO, you'd have to do the reverse of what aussie has said. Previously have your lays been coming 2nd and 3rd and have now just taken their turn at winning or were they previously unplaced and now started winning?

Also are there any logical extensions to your system that would rule out a portion of your losers without taking away any (or none) of your winners? Obviously adding a backfitted rule like must not have been within 2.3 to 3.7 lengths of the winner last start is not appropriate, but you may find something logical that fits with your system which rules out a number of the losses.

In saying that, i'm obviously not on the same level as some of these others guys but I think if you took a combination of UB's, CP's and Aussies idea's you'd find out if you need to keep it or can it.

I heard they finally located Binh Tran, and some time ago three guys were working on resurrecting it, but haven't heard anything for about a year or so.

The Ocho, laying horses 1 out, and your bank has lost over 40% from its peak should tell you something. I will elaborate on a new thread as I have to answer a old query from the Lord.