Personal Journey - discovering WHY many strategies don't work

[Fugazi]

How far above BSP to be sure you're making profit?

I've been analysing another failed method of mine (small sample, 30 bets). I have found that on cross referencing BSP and the odds I backed at, I seem to be above BSP 80% of the time!

so for example I back at 2.28 - BSP was 2.24

I know there is no strict answer, but I wonder when you consider commission and BSP not always being the true value, what people think a rough ballpark would be to profit long term? Also, I need to check the times I was under value, if these are wildly more under value than the times I got over value. I suspect this will be the most likely finding in a bigger sample.

Beating BSP is a very good starting point. I doubt that any of us can beat it 80% of the time over a reasonable sample.

You also need to consider the entire equation: 80% * (size of beat) + 20% * (size of miss)

As you are a beginner, if you're beating BSP by 80% you should get out of your trades at BSP and minimize your commission. It'd be very difficult to blow a bank while you learn.

I was thinking more of just backing at these prices and not trading (so really just pure value betting). The reason the method failed was that I wasn't able to trade out successfully everytime. It was an automated scalp sometimes it wasn't getting the bet to trade out fast enough (I need a bigger sample to see if I could start backing at best available price in these instances)

As you said, tiny sample size. But currently letting it run for a bigger sample to then try find a pattern from the best value bets, in the likely event 20% * (size of miss) is too large to profit currently.

[Anbell]

How far above BSP to be sure you're making profit?

I've been analysing another failed method of mine (small sample, 30 bets). I have found that on cross referencing BSP and the odds I backed at, I seem to be above BSP 80% of the time!

so for example I back at 2.28 - BSP was 2.24

I know there is no strict answer, but I wonder when you consider commission and BSP not always being the true value, what people think a rough ballpark would be to profit long term? Also, I need to check the times I was under value, if these are wildly more under value than the times I got over value. I suspect this will be the most likely finding in a bigger sample.

Beating BSP is a very good starting point. I doubt that any of us can beat it 80% of the time over a reasonable sample.

You also need to consider the entire equation: 80% * (size of beat) + 20% * (size of miss)

As you are a beginner, if you're beating BSP by 80% you should get out of your trades at BSP and minimize your commission. It'd be very difficult to blow a bank while you learn.

I was thinking more of just backing at these prices and not trading (so really just pure value betting). The reason the method failed was that I wasn't able to trade out successfully everytime. It was an automated scalp sometimes it wasn't getting the bet to trade out fast enough.

So I need to avoid the times it placed a lay bet but then no back bet.

As you said, tiny sample size. But currently letting it run for a bigger sample to then try find a pattern from the best value bets, in the likely event 20% * (size of miss) is too large to profit currently.

You can automate trading out at BSP thanks to Dallas as always viewtopic.php?f=58&t=15199

Trading out is almost always the best option if possible (if you're measuring your success by beating BSP), especially for a beginner

[Fugazi]

Beating BSP is a very good starting point. I doubt that any of us can beat it 80% of the time over a reasonable sample.

You also need to consider the entire equation: 80% * (size of beat) + 20% * (size of miss)

As you are a beginner, if you're beating BSP by 80% you should get out of your trades at BSP and minimize your commission. It'd be very difficult to blow a bank while you learn.

I was thinking more of just backing at these prices and not trading (so really just pure value betting). The reason the method failed was that I wasn't able to trade out successfully everytime. It was an automated scalp sometimes it wasn't getting the bet to trade out fast enough.

So I need to avoid the times it placed a lay bet but then no back bet.

As you said, tiny sample size. But currently letting it run for a bigger sample to then try find a pattern from the best value bets, in the likely event 20% * (size of miss) is too large to profit currently.

You can automate trading out at BSP thanks to Dallas as always viewtopic.php?f=58&t=15199

Trading out is almost always the best option if possible (if you're measuring your success by beating BSP), especially for a beginner

The issue is my lay bets may be above BSP , hence I need to wait for a bigger sample first to see if thats true then adjust the automation once I know that (seems the markets were actually telling me the opposite of what I thought when I created the automation)

What I thought was going to be a drift in price is actually near a rebound point

[conduirez]

I was thinking more of just backing at these prices and not trading (so really just pure value betting). The reason the method failed was that I wasn't able to trade out successfully everytime. It was an automated scalp sometimes it wasn't getting the bet to trade out fast enough.

So I need to avoid the times it placed a lay bet but then no back bet.

As you said, tiny sample size. But currently letting it run for a bigger sample to then try find a pattern from the best value bets, in the likely event 20% * (size of miss) is too large to profit currently.

You can automate trading out at BSP thanks to Dallas as always viewtopic.php?f=58&t=15199

Trading out is almost always the best option if possible (if you're measuring your success by beating BSP), especially for a beginner

The issue is my lay bets may be above BSP , hence I need to wait for a bigger sample first to see if thats true then adjust the automation once I know that (seems the markets were actually telling me the opposite of what I thought when I created the automation)

What I thought was going to be a drift in price is actually near a rebound point

If the trade has gone the wrong way, then trade out for a small loss, and then analyse why you lost, look at everything as to why the trade went wrong. Do not just take a punt, trade out.
You should look at this video, it explains a lot about trading.

https://www.youtube.com/watch?v=ImLrhC74pDw

[Fugazi]

You can automate trading out at BSP thanks to Dallas as always viewtopic.php?f=58&t=15199

Trading out is almost always the best option if possible (if you're measuring your success by beating BSP), especially for a beginner

The issue is my lay bets may be above BSP , hence I need to wait for a bigger sample first to see if thats true then adjust the automation once I know that (seems the markets were actually telling me the opposite of what I thought when I created the automation)

What I thought was going to be a drift in price is actually near a rebound point

If the trade has gone the wrong way, then trade out for a small loss, and then analyse why you lost, look at everything as to why the trade went wrong. Do not just take a punt, trade out.
You should look at this video, it explains a lot about trading.

https://www.youtube.com/watch?v=ImLrhC74pDw

What I am getting at, is I hypothesise both lay and back are above BSP, before price then rebounds. I have 6k profit from some value betting football 2ups in Bet365 (expecting to be gubbed soon) so am familiar with what a value bet is and have the discipline

I am not blindly punting, I believe I may be finding value that I am unable to successfully trade out. But I'm in the process of generating a bigger sample to check my hypothesis and analyse further

i'll check out the video though, sounds great!

[grimm]

The issue is my lay bets may be above BSP , hence I need to wait for a bigger sample first to see if thats true then adjust the automation once I know that (seems the markets were actually telling me the opposite of what I thought when I created the automation)

What I thought was going to be a drift in price is actually near a rebound point

If the trade has gone the wrong way, then trade out for a small loss, and then analyse why you lost, look at everything as to why the trade went wrong. Do not just take a punt, trade out.
You should look at this video, it explains a lot about trading.

https://www.youtube.com/watch?v=ImLrhC74pDw

What I am getting at, is I hypothesise both lay and back are above BSP, before price then rebounds. I have 6k profit from some value betting football 2ups in Bet365 (expecting to be gubbed soon) so am familiar with what a value bet is and have the discipline

I am not blindly punting, I believe I may be finding value that I am unable to successfully trade out. But I'm in the process of generating a bigger sample to check my hypothesis and analyse further

i'll check out the video though, sounds great!

When you say you not blindly punting what stats do you have. For example
All weather race stats, last 5 year fav win strike rate 33% . In this data I haven't check anything else just all weather races favt win % (odds between avg 2.5 to 5.5) . I can't blindly back favt after this stats. Then I add lets say trainer win%, jockey win %, horse form and class. Etc. Strike rate jump to 46% then you add some other stuff.
To save some money, find some data then do back test what you looking for.
Use filling gap like let say 4 tick below best price with off-set couple of tick and green up last 10 sec automation.
Hope that helps.

[Fugazi]

If the trade has gone the wrong way, then trade out for a small loss, and then analyse why you lost, look at everything as to why the trade went wrong. Do not just take a punt, trade out.
You should look at this video, it explains a lot about trading.

https://www.youtube.com/watch?v=ImLrhC74pDw

What I am getting at, is I hypothesise both lay and back are above BSP, before price then rebounds. I have 6k profit from some value betting football 2ups in Bet365 (expecting to be gubbed soon) so am familiar with what a value bet is and have the discipline

I am not blindly punting, I believe I may be finding value that I am unable to successfully trade out. But I'm in the process of generating a bigger sample to check my hypothesis and analyse further

i'll check out the video though, sounds great!

When you say you not blindly punting what stats do you have. For example
All weather race stats, last 5 year fav win strike rate 33% . In this data I haven't check anything else just all weather races favt win % (odds between avg 2.5 to 5.5) . I can't blindly back favt after this stats. Then I add lets say trainer win%, jockey win %, horse form and class. Etc. Strike rate jump to 46% then you add some other stuff.
To save some money, find some data then do back test what you looking for.
Use filling gap like let say 4 tick below best price with off-set couple of tick and green up last 10 sec automation.
Hope that helps.

Yes, I suppose I am rushing. Im betting, then analysing.

Rather than analysing, testing, re analysing.

[Anbell]

What I am getting at, is I hypothesise both lay and back are above BSP, before price then rebounds. I have 6k profit from some value betting football 2ups in Bet365 (expecting to be gubbed soon) so am familiar with what a value bet is and have the discipline

I am not blindly punting, I believe I may be finding value that I am unable to successfully trade out. But I'm in the process of generating a bigger sample to check my hypothesis and analyse further

i'll check out the video though, sounds great!

When you say you not blindly punting what stats do you have. For example
All weather race stats, last 5 year fav win strike rate 33% . In this data I haven't check anything else just all weather races favt win % (odds between avg 2.5 to 5.5) . I can't blindly back favt after this stats. Then I add lets say trainer win%, jockey win %, horse form and class. Etc. Strike rate jump to 46% then you add some other stuff.
To save some money, find some data then do back test what you looking for.
Use filling gap like let say 4 tick below best price with off-set couple of tick and green up last 10 sec automation.
Hope that helps.

Yes, I suppose I am rushing. Im betting, then analysing.

Rather than analysing, testing, re analysing.

Betting is ok, just bet small to collect data

[Fugazi]

When you say you not blindly punting what stats do you have. For example
All weather race stats, last 5 year fav win strike rate 33% . In this data I haven't check anything else just all weather races favt win % (odds between avg 2.5 to 5.5) . I can't blindly back favt after this stats. Then I add lets say trainer win%, jockey win %, horse form and class. Etc. Strike rate jump to 46% then you add some other stuff.
To save some money, find some data then do back test what you looking for.
Use filling gap like let say 4 tick below best price with off-set couple of tick and green up last 10 sec automation.
Hope that helps.

Yes, I suppose I am rushing. Im betting, then analysing.

Rather than analysing, testing, re analysing.

Betting is ok, just bet small to collect data

Yeah I only do £1, doesn't cost a lot as some win and I find it more useful and engaging than practice mode

[Fugazi]

Rollercoaster of emotions today. Logged in after work after leaving my automation placing £1 bets. To my surprise, nicely profitable! SADLY, on inspection, I was well below BSP on the whole. Excluded the AUS races as Timeform didn't have BSP. I know there are easier ways than manually going to Timeform, but haven't got round to that yet. More exciting just having a stab at the moment.

HOWEVER. This small sample has given me some new information to investigate:

A) laying on this small sample would be value long run, including 2% commission, and adjusting tick to lay at the same price im backing
B) Had I reduced the start time of the automation pre race , backing would be profitable.


So I now have two new ideas to try out with larger samples. AND I got lucky with a losing strategy today

[Fugazi]

Still small sample, but growing in confidence I'm close to cracking this. Even if this small sample is just lucky variance, then I'm confident a few tweaks and I've got it. Im shocked at how much info is actually given away for free in this forum when you figure out whos advice to listen to and piece the clues together.

I am actually just as profitable (ROI) over double this sample size the last 10 days or so, but I ruined it with some none working strategies hah.

[NotBothered]

I will give you my 2 cents

First off no automation listed here on this forum will make one cent as is -- and even the concepts themselves are flawed -- ie you may make a few winning trades for small profits but you will get wiped out by large losses.

So then you realize that each strategy needs a certain set of markets to work in - and next to match a strategy to a market.

Then someone will say well you need to stop loss quickly - OK that removes a tonne of markets because thin markets move quickly and in low liquidity.

All of these things you have pointed out.

WOM - price movements are a funny thing -- alot of the money flow is spoofing - and even I do this myself - I back one thing to get a better price on another at times - or place larger bets that will influence the market.

What has been my Solution to this ?

Buiild a Database - identify the trends and flow - and stats and transfer that into strategy -- where you control -- race selection - price and SR .. this allows alot more chance of success.

+ anyone can make a video of one trade and make a profit -- but it is doing the same thing 100 times that really makes a difference