Wednesday, 8 April 2009

lay the field - my favourite racing strategy



Dabbling with laying the field in-running at various prices today, not just one price, but several in the same race. Got several matched in the previous race at Brighton, then this race came along at Nottingham. Such a long straight at Nottingham makes punters often over-react and think the finish line is closer than it actually is. As you can see by the number of bets matched, there was plenty of volatility in this in-play market. It's rare you'll get a complete wipe-out with one horse getting matched at all levels, but it can happen, so don't give yourself too much risk...

30 comments:

  1. Hi Scott,

    nice green figures! You are using this kind of trading on many sports(horses, MotoGP, biathlon).
    May I ask how much was your maximum liability? And do you take a stop loss, when the odds move against you?

    Cheers, Loocie

    ReplyDelete
  2. Sorry Scott - I don't understand what you exactly did and what is "laying the field". Can you please provide more explamations.

    Best regards
    Rotteborp

    ReplyDelete
  3. Hi guys,

    Here's the strategy - http://betting.betfair.com/education/4-horse-racing/07-lay-the-field/trade-lay-all.html

    When you lay two selections at odds-on in any (one winner) event, a tennis match, football etc, you make profit, so long as you do it for level stakes. You can also use the same principle for laying 3 horses at less than 3, 4 golfers at less than 4 etc.

    My worst case liability in this race was about £20 if one horse hit the front and won easily. But that rarely happens, at least when offering prices as high as 3.35 and 2.32, you will at least get some matched.

    All I've done is repeat the strategy in the article (look at the video clip linked at the end for further clarification) several times in the one race.

    ReplyDelete
  4. that link from my last post has changed, it's now:

    http://betting.betfair.com/education/4-horse-racing/07-lay-the-field/trade-lay-all-101108.html

    or go straight to the video clip at http://www.screencast.com/t/dgtlUaWN26

    ReplyDelete
  5. Hi Scott

    Just to let you know that it was the Oscars night last night and unfortunately you did not win, but I believe this post has won you even greater acclaim !!!

    Browse over to this link to see what I mean.

    http://patientspeculation.blogspot.com/2011/02/my-blog-post-of-week-week-two-winner.html

    I hope you approve.

    Mark

    ReplyDelete
  6. Bingo!!
    Hello Scott
    I have just stumbled across your site from a google search. You seem to be the only person so far that understands this method of lay all that I want to use.

    Do you know of any bot that can actually automate this method of 2 horse under 2/1, 3 under 3/1 etc. I need something simple as opposed to Gruss I have on trial and noone seems to know how to write the correct formulas for this.

    Any thoughts/suggestions would be appreciated by any of us here

    regards
    shane

    ReplyDelete
  7. Hi Shane,
    None that I am aware of, I've always done it manually and preferred to vary it myself. Also, if you get in early, your price will get slashed by non-runners. A decent coder could write it for you, but I've never chased one.

    Try the Gruss forum perhaps - someone in there should know how to write one for you.

    Cheers
    Scott

    ReplyDelete
  8. just found this site and have to say it's excellent.....been looked at the lay the field video and can't understand why the screen which initially shows a liability of £5 per horse suddenly changes to £12.46 at the off. Surely the liability remains at £5....is this an error on the video ?

    ReplyDelete
  9. Hi Nigel - if you look closely on the top right of the bet slip, the setting switches from 'Liability' (£5) to 'Payout' (£5 risk + £7.46 backer's stake). The bet slip switches between the two when the unmatched bets are switched to Keep.

    ReplyDelete
  10. Thanks Scott....have been keeping an eye on laying the field in handicaps at the aw meetings..seems to be profitable at Southwell and Wolver but not at Kempton and Lingfield.Very early days yet but think this has the making of profitable system.Am using 1.67 but have already reduced the lay to 1.26 on the fav. Hope to eventually arive at optimum laying prices.

    ReplyDelete
  11. Hi Scott, may I ask you how about any the exit strategy here? Or do I just have to leave this running and hope for the best and trust the selection of the race and the right prices layed? The price fluctuations are crazy and I was trying to hedge at some times using the toy but it seems hard to get matched at all. is an green hedge favorable, if only one horse hit the lay price and drifted... or should I just be patient. I tested on training mode with 10 euro liablity and two times already occured that the favorite got matched drifted out so i could easily green for 5 to 7 euro but did not do it and the fav got back to win and was the only lay being matched. I will not try to investigate as much as possible in training mode to see if this is working out. the only problem is that the toy did show me a loss although i had like 6 or 7 horses matched at 3 different lay stages and all the stages had at least 2 matched. looking at pay loss live figure on betfair website typing in the same amounts at prices into the betslip, it turned out to be 28 euro profit. strange.. hope this was only a software mistake.
    cheers!

    ReplyDelete
  12. TTS - it's a set and leave strategy, and it will never happen on every race. Some races are won by horses who lead all the way and never look like losing. Other races are complete carnage. There are arguments for trading out or even just cancelling the rest of the bets after you get one matched, but that's not how I do it.

    If using the Toy (I don't use it so can't comment on its features and faults), I would log into BF in the background to keep an eye on it. If you get multiple horses matched across several lines, then you must be in profit (unless you have mucked up the liability stage - should keep them fairly similar).

    There are absolutely no guarantees that this strategy will work in the long run. Like any system it will have good and bad runs. Worst case scenario long-term, you will lose a few %. I am convinced that if you tinkered with it - got your pricing points right, varied them by track, distance, price of favourite, field size, even track condition, that you could profit in the long run with it.

    It's a bit of fun designed to show you the volatility of Betfair markets in a short period of time, teach the percentages of laying, and get people involved in a sport which not everyone is interested in, because those markets happen every few minutes. A bad run will have small losses adding up slowly, but a great run will have jackpot days - eg when three horses are beaten at 1.03 on the same day, four matched at 1.31 in the same race etc.

    ReplyDelete
  13. Wow, that reply was fast. Thanks for the info... I will just try my best and investigate the tracks, number of horses and prices to lay at. right now I use 3 different stages at 1.8, 2.7 and 3.6 halving the liability at each stage. Just watched your video when the money is on the favorite, looks decent too, so i will keep an eye on these markets as well. thanks for the great info

    ReplyDelete
  14. those are very high price points - at 3.6, you need nearly 4 horses matched to make a profit (2 matched below 2, 3 matched below 3 etc). This system is all about LOW liability, BIG potential upside - my lines are something like 1.67, 1.35 and 1.05. In big fields, I might expand above the 2.0 mark, say 2.34 and 3.2. But when you are above 2.0, you are risking more than you stand to win - that's dangerous and completely against the principle. Remember this is about learning as much as winning, thus the risk must be kept low.

    ReplyDelete
    Replies
    1. Hi Scott,

      Saw in previous posts that you usually lay at 3 different odds 1.67, 1.35 and 1.05.
      Is there an advantage on laying 3 odds instead of a single laying at for eg. 1.8?
      Do you use different stakes for each?
      Have you done any king of backtest to see at what tracks is lay the field more profitable?
      Thanks,
      Hugo

      Delete
    2. Hi Hugo,

      Multi-lines is to cover more bases. I play this method to land the jackpot occasionally, i.e. when a 1.03 gets turned over, hence multiple of lines. There were five lines in this race. Depends on the type of race, size of field, price of the favourite, distance, number of runners, particular track...

      Stakes differ, liability tends to be slightly lower at low prices (but potential win much higher).

      Firstly, before talking about tracks and criteria, it's important to realise Betfair markets evolve over the years. What was a winning strategy five years ago may not be now. The margin of edge will certainly have decreased. What worked 10 yrs definitely won't work now.

      On a bit of research through the BF historic data, the best tracks were:
      (NH)
      Worcester
      Towcester
      Huntingdon

      (Flat)
      Yarmouth
      Leicester
      Doncaster

      Reasons - camera angles are important. You want tracks where it is tough to work out who is in front, how close the horse making a late run really is, etc. Those tracks either run at the camera (Worcester), have long and wide straight tracks where horses often split down both sides of the track (flat), or have a nasty hill which is deceptive to the inexperienced jockey/poorly performed horse (Towcester).

      Distances - remember, you need long enough for other punters to react. A 25 horse dash over 5f down a straight will rarely work as well as you think - it takes too long for punters to isolate a horse who might win. When they do, that one invariably wins. I found my best results started at 7f.

      Field sizes - when the fields are too big, it's tough for one horse to shorten up as much as you need (as others are still a chance in the race, and in theory, the market % should be near 100). This is where I often post higher lines, looking to get three horses matched <3 rather than two at 1.3.

      Price of favourite - relates to field size as well. If its a 16 runner handicap with 8/1 the field, it will be very difficult for one horse to shorten up enough to be shorter than he should be. Remember, you are trying to lay over-reactions, not true prices.

      I also found results seemed to be better around change of season - flat to jumps & vice-versa, plus when the weather turned around sharply - wet tracks after long periods of firm ones, and vice-versa.

      Best advice I can give is try it for yourself and see what you find. It's a long term game so don't count anything as meaningful without a few hundred races under your belt. You can use the Timeform/Betfair form site to look at historical prices traded too.

      Delete
  15. Very interesting post, I have been trying a similar method in Golf tournaments, where I have been laying the field in the 'Top 5 Finish' market.

    So far I have had mixed results, but in some of the tournaments the liquidity has been very poor, and as you say Scott, it's more about tinkering and finding out what 'lay points' work in the long run.

    Will stick with it anyway, was laying at 3, but have dropped down to 2 for 'The Players Championship,' where the liquidity is much higher than the previous tournaments I have tried it at.

    ReplyDelete
  16. multi-winner events - i.e. place markets, are not ones I've tried this out on, but I would definitely be going lower in price, as at 2.0 you need 10 players to be matched to break even (500% market with five winners, thus 10 x 50% required, where 2.0 is equivalent to 50%), plus you have dead-heats to worry about. Initial thoughts are you'd have to pick your events wisely as regular events may not see enough volatility and liquidity.

    ReplyDelete
  17. Hi Scott
    Really like this strategy and have used it with modest sucess -albeit small stakes.
    I've been using some research from betfair data to try and help me pick suitable races
    -type of race, average low in run price, BSP etc.
    Just a quick question. You mention using odds of 1.67 , 1.35, & 1.05. Lets assume £10 level stakes
    that make a liability of £10.70 on the eventual winner , so the second will have to match the first two lines
    to make a profit. Do you think there is any benefit in just having two lines but wider apart say. 1.11 & 1.81
    giving you slightly more chance of catching runners that stray over evens. Just a thought.

    Excellent site, really enjoy it.

    Cheers mate

    Gareth

    ReplyDelete
  18. Hi Gareth,

    I think the best way to vary the system is by race type, venue, field size, price of fav, quality of camera angle, distance, conditions etc. Some courses are more prone to 1.05s being turned over, others rarely see any horse beaten below 1.7 (or more correctly, don't see any more than you would reasonably expect).

    At 1.81, with two matched and not reaching the lower line, you are only just breaking even. But with that comes more chance of getting three, four or even more matched.

    ReplyDelete
  19. Hi Scott
    Thanks for prompt reply - tried my way today. both races two matched 'but they would also have hit the middle lay - If of course I'd had one !!! So no loss but down on profit.
    So I think having tried it it's back to your way !! Of course you know what will happen tomorrow!!!!
    Apart from what you've said above, any further advice re: race selection- which lets face it is the key to this strategy, would be very welcome.

    Cheers mate

    Gareth

    ReplyDelete
  20. Gareth,
    See reply to Hugo earlier in the comments.

    Cheers
    Scott

    ReplyDelete
  21. Scott
    Thanks for not saying I told you so !!
    Just one more -I notice on the screen shots at the top of the page that you've put up offers above evens, with various different stakes - I appreciate that you vary the odds depending on things like field size but do you have any advice on tailoring the stakes to the price -is there a clever formula ?
    Incidentally I mentioned earlier that I have been using some betfair data to assist, so for anyone interested they might like to try this link
    http://www.betfairpromo.com/betfairsp/prices/index.php
    Hope someone finds it as useful as I do.
    Thanks for your time mate

    Cheers
    Gareth

    ReplyDelete
  22. No science, just be careful not to risk too much on prices that high as the risk is greater than the reward at odds-against. And while it might sounds like a great idea in a big field... the bigger the field, the harder it is for prices to move in an overreaction. Remember, 'true' prices should let you fare no better than breaking even in the long run, it's the overreactions which make you the money.

    ReplyDelete
  23. A few thoughts/experiences on this subject:

    1. I built up a database for about two thirds of courses with all results from the start of 2008 onwards. This date was selected since the BF crossmatcher came in at about March 08, and some horses may not go as low IR as they did before its inception. For each course I looked at the ROI laying the field for all prices below 2.0 ; a couple of examples of the type of results can be shown graphically eg here are the plots for Market Rasen (near 700 races) and Wincanton (just over 500)

    http://i572.photobucket.com/albums/ss164/Sheva_04/Horses/MRupdate_zps5f9ea7f0.jpg

    http://i572.photobucket.com/albums/ss164/Sheva_04/Horses/Wincupdate_zpsf361bd5b.jpg

    X axis shows the price at which the field is laid, and Y axis the resulting net ROI (assuming 5% commission).

    These 2 tracks are at roughly extreme ends for the courses I’ve looked at. At Wincanton you’re always going to lose 20% or more of your money in the long term whatever price you lay at … due to course characteristics, race tactics, camera angles etc. You could argue that by concentrating on a given subset of race type or distance, results will give a better outcome, but when the overall trend for a course is so negative, i’m wary of trying to find a race type that bucks it.

    Conversely, Market Rasen shows that you’d have made a bit of money laying the field in every single race in the 1.5 to 1.7 area, and also at much lower prices sub 1.2. At this point, further exploration by race type or distance to identify even more rewarding returns can be a lot more of a useful tool, since it’s based on an overall trend which seems fundamentally sound over a large sample of a few hundred races.

    2. Price(s) at which to lay the field:
    When I started out I was trying to pick more than one lay price, but now I just choose the one which will give the theoretical return based on previous data. Previously, when I felt the best price was quite a low one, maybe 1.3 or less, the temptation was to also put in additional lays at higher prices to guarantee some sort of return and avoid too much variance in non-winning races. However, over a large number of races, choosing what you believe to a single best price (on whatever basis you use) is IMO the way forward.

    3. Effects of ground:
    Filtering results by soft ground and fast ground, I don’t think ground conditions are of any relevance in the long term. One or two courses do show a slight preference for bad (soft or worse), but likewise for a few courses the opposite is true. Over 10,000+ races (flat and NH), the ROIs for ‘good or better’ and ’good to soft or worse’ are pretty much the same. There seems to be a perception that heavier conditions lead to more horses being beaten at lower odds IR, but the reality is that the field often gets well strung out when this happens, so a fair share of winners are in fact identifiable a long way out .

    4. Flat vs Jumps
    Based on the last couple of years’ data, I think the strategy is becoming a better one to use for the flat rather than over the jumps. My guess for this would be the whip rules that have been established in that period and its possible greater impact in the national hunt sphere: horses dropping away quicker meaning easier identification of winner.

    Sheva.

    ReplyDelete
  24. Brilliant stuff there Sheva. One query re ground conditions - I especially like to use this method when track conditions change after a long weather pattern - first week of dry ground after a month of wet, and vice-versa. Are you able to test that at all?

    ReplyDelete
  25. Very interesting, sheva04. I've got similar data from odds-on IR laying. I've only got data from 2009 to hand, but it's pleasing to see that we get the same results!

    From 510 races at Market Rasen, 1.03 looks like the most profitable odds (1.01 and 1.02 may be better but my calculations assume you will be some way back in the queue and need a lot more money traded before you can be confident of being matched at these low competitive odds). Most < 1.15 odds look good, then suddenly odds around 1.56 are also good. So basically my data matched your graph.

    From 394 races at Wincaton, I also find all are loss-making, with 1.72 the 'best' and 1.01 the worst. Again, looks roughly like your graph!

    Interesting if you just take races from 2012 onwards, the results are broadly the same for both tracks.

    n.b. my results have lots of assumptions, like how far back in the queue you will be to get matched, they are based on laying to risk a fixed liability regardless of odds, my own commission rate, and they also don't take account of short-priced favourites. (e.g. when testing laying the field at 1.5, if there's a 1.2 shot in the race, it blindly assumes you lay it at 1.5 regardless. Heavy odds-on favs are rare, but this still might bias against tracks that have more short priced favourites?)

    Tracks that seem to be terrible for laying odds-on IR at any odds are: Ascot, Ayr, Bath, Kempton, Lingfield, Musselburh, Plumpton, Pontefract, Wincanton, Wolverhampton, Worcester and Yarmouth. Note that most of the A/W tracks are there! I suspect those busy tracks are most popular for on-course IR players who are going to snap up all 'sure' winners before the line. Perhaps for those races, blindly following the money and backing the leader might be profitable. I published a program that did just that, but that was a long time ago...

    Anyway, I've not done any research into comparing profits v.s. going. However, I suppose it wouldn't be too hard to do. Scott, how do you find these changed weather conditions? Do you mean something like spotting a meeting where the declared going is markedly different from the previous track meeting's going, or would that be too simple? I could investigate something simple like that, but I've got limited historical track weather condition data to do anything more clever.

    One final thing: The risk when you narrow down these stats is that your data set just becomes too small and conclusions will soon turn to noise. Per-track odds are probably valid, given enough races (but you still have to assume that the betting patterns aren't changing too much over time, like more/less liquidity over time, etc). However, good luck if you want to analyse IR at Laytown only when the weather has changed :)

    ReplyDelete
  26. Thanks 'unknown'. For that list of tracks, I'd suggest the A/W tracks plus Ascot have camera angles that are too good, so anything close to an edge gets snapped up by the on-course IR players.

    Worcester and Yarmouth - really? I'd have them at the top of my list of courses to play. Perhaps I still remember the good old days of 2006-7 and the market has evolved - i.e. other layers have found it and the value is long gone.

    Re changed weather conditions - in very simple terms, April and October are usually the months for it, but with UK weather, it can happen any time of year.

    ReplyDelete
  27. (no idea why my name appears as 'unknown' when using Google sign in!)

    My list of bad tracks was based on a range of dates, mostly more recent than the 2009 data. Sorry, that was a bit confusing!

    Worcester used to be good. From 2009 onwards (462 races), laying around 1.01-1.03 was great, and 1.33 also good. It has slowly gotten worse and from around 2012 looks unprofitable. Yarmouth I can't see any profits whatever year I pick for the start date!)

    Here's a sample of the raw data, to show how the track seems to be getting worse: (might just be mangled by the formatting):

    Profit for track Worcester (from 462 events, 2009-)
    1.02 1.03 1.05 1.06 1.33 1.34 1.31 1.01 1.35 1.25
    4.82 4.03 2.49 1.83 1.51 1.49 1.40 1.35 1.28 1.20
    15 18 29 33 159 164 151 6 166 122

    Profit for track Worcester (from 229 events, 2011-):
    1.40 1.37 1.35 1.53 1.33 1.52 1.41 1.34 1.36 1.57
    0.27 0.13 -0.02 -0.03 -0.07 -0.08 -0.08 -0.08 -0.13 -0.23
    88 80 74 111 69 110 88 72 76 117

    Profit for track Worcester (from 109 events, 2012-):
    1.37 1.57 1.33 1.53 1.36 1.52 1.38 1.40 1.54 1.58
    -0.84 -0.98 -1.00 -1.03 -1.13 -1.15 -1.15 -1.16 -1.20 -1.21
    37 55 32 51 35 50 37 39 51 55

    First row is the odds, sorted from most profitable, left to right.
    Second row is average per-race profit (based on assumed liability, commission, position in queue etc. Best to just treat as a relative comparison. Should be proportionate to ROI)
    Third row is number of winning lays at those odds.

    YMMV. Completely different results might be achieved with slight tweaks to the assumptions, especially stake size.







    ReplyDelete
  28. I haven’t done Worcester, since a lot of recent meetings have been lost due to flooding issues and may well continue. Maybe this could be a reason for recent results, with trainers having less confidence it state of very variable ground, and therefore reduced field sizes or less competitive races (highly speculative).

    Likewise I see very few redeeming features in Yarmouth, graph below:

    http://i572.photobucket.com/albums/ss164/Sheva_04/Horses/Yarmouthnew_zps102865e6.jpg

    But the fact that it may have been a good course a few years back does raise the issue of how many races are required to have confidence in the whole methodology and belief that there’s some sort of real edge there if you dig deep enough. Performing it on 4 or 5 years data is still quite transient.

    ReplyDelete

Thanks for your comments, but if you're a spammer, you've just wasted your time - it won't get posted.