Sports Day Trading

[Data Miner]

Friday 26 January 2007

1 win from 1 yesterday adding +90 points to January pot, carry forward now standing at 267.96pts, which is nicely in front of my 250 point monthly target.

Two trades for me today at Hereford, selling both SPs and Favourites



Happy Spreading

 

[Data Miner]

Friday 26 January 2007 - supplementary

All was going very well up to the penultimate race, day ended with a 23.1 point loss. However, reconciliation of my spreadsheet records with the spread statements went in my favour (must stop checking the make-ups with IGIndex, they're all too often inaccurate) so carry forward is now +261.05

I'm not going to be around for Saturday's racing so I'll be giving it a miss, but do have 2 football mini-performances in place;

BUY Notts County
BUY Stafford Rangers



Happy Spreading

 

[Data Miner]

Sunday 28 January 2007

1 win, 1 lose on the minis reduces the carry forward by 9pts to 252.05, only just ahead of my January target.

Two racing bets today, both at Kelso. Selling Favs for 2pts, selling distances for 3pts.


Happy Spreading

 

[Data Miner]

Monday 29 January 2007

2 from 2 yesterday ups the January carry-forward by 65.95 pts to +318

Just a single trade today, sold Ludlow distances @ 33
(which, after only 2 of the 7 races completed is currently 14.5 . . . so not looking too good)


Happy Spreading

 

[Data Miner]

Tuesday 30 January 2007

Small reversal yesterday, knocking 7.35 pts off the bank, reduced now to +310.65

One distance sell at Taunton today. No mini-performances tonight though.



Happy Spreading

 

[a_gnome]

Hi Data Miner,
Interesting thread - I do a bit of scalping on BetFair myself.

I didn't read your previous thread so am not familiar with what you are doing. Can you elaborate a bit on why you pick certain trades? Are you looking to get certain prices on these trades when you do them? Do you have some algorithm for determining good prices or is it more seat-of-the-pants type of stuff?

I'd be interested to hear more about what you are doing if you'd care to divulge.

 

[Data Miner]

a_gnome,

Past racing results processed using artificial neural network, applied to daily horseracing spreads.
If the NNet prediction has a sufficient edge over the quotes, a trade is triggered.

No seat of pants stuff at all - entirely emotion-free statistics driven. Personal skill, judgement and/or emotional trading for me always ends in tears.

 

[Data Miner]

Wednesday 31 January 2007 - Final day

The single bet yesterday was successful adding another +71.25 to the carry forward. January bank now standing at +381.90

Two racing spreads today;
SELL distances at Newcastle for 3pts
BUY favourites at Newcastle for 2.5 pts



Happy Spreading

 

[a_gnome]

a_gnome,

Past racing results processed using artificial neural network, applied to daily horseracing spreads.
If the NNet prediction has a sufficient edge over the quotes, a trade is triggered.

No seat of pants stuff at all - entirely emotion-free statistics driven. Personal skill, judgement and/or emotional trading for me always ends in tears.

DM
Very interesting and high tech! How long did it take you to develop this? Can I ask what sort of factors go into your NNet? Of course if you'd rather not say that's fine but I'm just curious as to what inputs your network thinks are predictive and useful.

a_g

 

[Data Miner]

DM
Very interesting and high tech! How long did it take you to develop this? Can I ask what sort of factors go into your NNet? Of course if you'd rather not say that's fine but I'm just curious as to what inputs your network thinks are predictive and useful.

a_g

Because UK racecourses are many and varied, RH, LH, figure-of-eight, level, uphill, downhill, undulating, exposed, sheltered, long straights, no straights, etc., etc. I therefore train nets for each venue individually. This, as you can imagine, is time consuming. But IMO well worth the initial effort.

The markets I trade are cumulative over a meeting and include winning margins (distances), odds of winners (SPs) and performance of race favourites.

In my experience the overwhelmingly single biggest influence on the above markets is the odds of each race favourite. But of course, this fact cannot be known with complete accuracy until each race has started However, the advent of Betfair now allows us a clearer pre-race picture than ever before

Other influential factors (to a greater or lesser degree, depending upon which course)

Ground conditions
Number of runners
Pro jockeys or apprentices/amatuers
Handicap race or not
Race distance
Flat race/ hurdle/ steeplechase

HTH

DataMiner

 

[Data Miner]

January 2007 Final Accounting

Final day brought with it 2 losers from 2 bets, reducing the monthly gain by -67.95 to +313.95
Still a satisfactory conclusion to January though, exceeding the 250 point monthly target.

A good start to 2007


Happy Spreading

 

[Helenqu]

My racing system is quiet at the moment too Too many meets abandoned. I've never used a Spreadbetting service for racing, I just use Betfair. What are the advantages of spread betting? Who do you use and why?

 

[Data Miner]

My racing system is quiet at the moment too Too many meets abandoned.

Oh! Sorry. My spread trading is accounted one month at a time. The thread followed my progress through January. Since then I've set up a forum on my own web site specifically for sports spread betting and machine learning. All of my daily trades will be recorded on there from now on.
However, I may reproduce the odd month of trading through this forum too from time to time.

I've never used a Spreadbetting service for racing, I just use Betfair. What are the advantages of spread betting? Who do you use and why?

Not sure there are any advantages of spread betting over odds betting, or vice-versa. It is whatever you're comfortable with. Personally, I find I can grind out regular profits via the spreads more easily than I ever did with regular odds betting or Betfair.

Hope that helps . . . I will be back! (sometime)

 

[Data Miner]

. . . I will be back! (sometime)

Hello again everybody!

Two full months of my sports spread trading have elapsed since I shared my +313.95 January gain through this thread. Here to report just how these two accounting periods have fared with regard to my aim of +250 points per month.

FEBRUARY TRADING
Through February I had just 19 trading days and struggled to achieve any sort of momentum until 10 days had passed, at which stage my persistence brought its rewards and a few good days ensured I made my target (and more). Ended the month with a +396.81 profit.

As the progress graph shows, a pretty steady month once the winners kicked in.

Anyone following my progress on a daily basis, maybe wishing to get an insight into the highs & lows of spread betting would have witnessed the full spectrum of emotions during the next full month of trading . . .

MARCH TRADING
The very first trade in March was a winning one, always nice to begin a new period with a positive return. However, after that things went rapidly downhill. A very soggy day at Newcastle racecourse cost me very dearly indeed ~ I went short on winning distances (amongst other things) and that single day saw my bank suffer to the tune of 856 points.

Now I use betting/trading "points" as a standard & easy way of illustrating stakes. So, at say, £10 per point, the above represented over eight-and-a-half thousand pounds DAILY loss.
At £100 per point, a £85,600 loss on the day.

As often is the case, things can often get worse before they get better. March it seems wanted to make that very point to me ~ maybe testing my resolve? Anyway, the losses finally bottomed out on day 9 at -1310.04.

Warning & Reminder!
A reminder to those thinking of dabbling in the spread betting arena ~ don't let me put you off, I think it's a great way of trading, but be thoroughly prepared as to what to expect as part 'n' parcel of the whole scene. The £10 per point trader in my situation was over £13k down after just 9 days of trading. To be a spread trader make sure you can not only cope with such a reversal, but putting new money down the very next day following the very same tactics you've used all month!

In my case resilience paid off, the position had been fully recovered by day 16, and even breaking +500 profit by day 20. March ended with a final day flourish (a 764.40 win on the final day) to end the month with my +250 target once again achieved comfortably at +877.84.

The graph for March clearly illustrates the roller-coaster ride!

Happy spreading!

 

[Data Miner]

I've been asked in a private email to give a brief explanation of the markets I trade. Rather than simply reply directly to that request, I'll also post the response on this thread so as to add a little flesh to the bones of what sports spread trading is all about. The sports spread firms of Sporting Index, SpreadEx & IGSport all have thorough explanations on their respective web sites of every market they offer. So just a brief run-down of the horse racing spreads I play;

Winning distances: (some firms refer to these as winning lengths ~ the margin of winning in horse racing is given in [horse] lengths)
Simply the aggregate winning margins (between winner & second placed) for each race over an entire meeting (usually six to eight races)
Jump (National Hunt) winning distances are, by nature of racing over greater distances - and with obstacles - usually greater by a factor of 4 or 5 than their flat equivalents. The spread firms have a ceiling of 30 lengths for jump races, this cap is set at 12 lengths for flat racing. (If the winning margin is greater - or maybe only one horse finishes!?! the upper limit is applied) As a concequence my staking for flat distances is higher than for jump racing.

Starting Prices: (SPs)
The aggregate of the odds of all race winners through a full meeting. 2/1 winner =2, 6/4 winner =1.5, 1/2 (ODDS-ON) winner = 0.5
The upper limit on SPs is 100/1, recently upped from 50/1 and is the same for all racing codes. The new limit gives this market the potential to be EXTREMELY volatile ~ one winner at 100/1 and an SP seller's hopes are in ruins from a single race, almost inevitably busting his position at a stroke. Every other winner at the meeting in such cases is taking the spread make-up further and further away, and likely costing the spread player dearly.

Favourites: (Favs)
Race favs in each race are awarded 25pts for winning, 10pts for a second place and 5pts for coming in third. Make-up is the aggregate of this points system throughout the meeting. Same set up for all racing codes.

Working this in practice? Coincidentally yesterday (Monday 02 April) I played in each of my three specialist markets;

1] SOLD Kelso distances @ 34 (IG) 5pts
2] SOLD Kelso SPs @ 59 (SX) 3pts
3] SOLD Lingfield favourites @ 74 (SF) 3pts

Results: (Make-ups)
1] distances MU=34.05 (LOSE) My return [diff between MU and my position] multiplied by stake = -0.25 point
2] SPs MU=39.9 (WIN) My return [Diff multiplied by stake] = +57.30 pts
3] Favs MU=85 (LOSE) My return = 11 * 3 = -33 pts

Day's return = +24.05

PS -Spread exchange
The Lingfield Fav sell was with Cantor Spreadfair, a firm not mentioned in my earlier reference. Spreadfair are unique in sports spreads inasmuch they operate as a trading exchange, spreads widths are often much tighter with SF, but commissions are changed against winnings (losing trades do not attract commissions) Commission starts at 5%, but is reduced on a sliding scale related to the level of financial activity in your account. Minimum commission is 3%, to achieve this however an account holder would need to be turning over (all profits & losses) of close on £200k per month.

PPS - What use tighter spreads?
Spreadfair (and a good few other spread firms, including those covering financial markets) make great play of their reduced spread widths, as though this is a boon to the spread player. The way I see it is ... What use to me is a reduced spread width??
If I'm looking to sell a position and one firm offer 54-59, another go 53-55, and a third have 55-65. It seems to me that comparing spread-widths is not going to improve my position, the obvious place for me trade is with firm [3] who offer me the best selling price.
The only time I can see it being useful is if the spread-bettor is playing a unique, one-firm market, and wants to retain the option of trading out of his/her position early. But I do not play unique markets (I prefer to be able to compare quotes) ... and NEVER close my position early. What are the views of other spread players?

PPPS
Football spreads are also a part of my sports spread regime, but I'll leave my run down of these markets until later.

Happy Spreading

 

[Data Miner]

03 April 2007
+186.95 b/fwd

Today's racing;
8] SOLD Wetherby SPs @ 53 (IG,SX) 3pts
9] SOLD Southwell Favs @ 67 (SX) 3pts

=====Previous trade (02 April 2007)=====
+162.80 b/fwd

5] SOLD Kelso distances @ 34 (IG) 5pts MU=34.05, -0.25
6] SOLD Kelso SPs @ 59 (SX) 3pts MU=39.9, +57.30
7] SOLD Lingfield favourites @ 74 (SF) 3pts MU=85, -66.00

Day's return +24.05, +186.95 to c/fwd

 

[NotQuiteRandom]

Hi Data Miner, my background is in financials rather than sport and I have never watched a football match, horse race or similar event. I am interested however in the term structure of the BetFair data which, to borrow an options analogy, exhibits a theta type property; though seemingly with a more linear structure through time.

In the case of football, as the match progresses the (BetFair traded) odds of a non-event occurring narrow in a predictable fashion (unless of course the event occurs). In selling these events one is taking a kind of event risk analogous to credit or some other binary event risk. See below for the data collected on a set number of goals occurring in a sample match. Data runs down from match end to match start.

1.01
1.02
1.03
1.04
1.05
1.07
1.08
1.09
1.1
1.11
1.12
1.13
1.14
1.15
1.17
1.19
1.2
1.21
1.28
1.3
1.31
1.32
1.33

My question therefore is how does one begin to model event probability in this context. I know plenty of models for examining credit risk, market jumps and other binary events but don't know how one would apply these to sport. Would it be appropriate to do so? Is there a branch of study in this area and can you point me in the direction of any information, papers or other resources online?

Likewise, I see a similar term structure in horse racing data. Below is a random sample from BetFair which has similar properties.

1.31
1.33
1.34
1.35
1.37
1.4
1.43
1.45
1.5
1.51
1.52
1.54
1.55
1.6
1.61
1.62
1.69
1.7
1.72
1.73
1.8
1.82
1.9
1.95
1.98
2.02
2.04
2.06
2.1
2.12
2.14
2.14
2.16
2.18
2.19
2.2
2.2
2.22
2.22
2.24
2.24
2.26
2.27
2.28
2.29
2.3
2.3
2.31
2.32
2.32
2.33
2.34
2.34
2.35
2.36
2.37
2.38
2.39
2.4
2.4
2.41
2.42
2.42
2.44
2.44
2.46
2.46
2.48
2.5
2.5
2.52
2.52
2.54
2.56
2.56
2.58
2.6
2.6
2.62
2.62
2.64
2.66
2.66
2.67
2.68
2.69
2.7
2.7
2.71
2.72
2.74
2.76
2.77
2.78
2.8
2.81
2.82
2.83
2.84
2.86
2.88
2.89
2.9
2.92
2.92
2.93
2.94
2.94
2.96
2.98
2.98


Thanks,

NQR

 

[Data Miner]

Hi Data Miner, my background is in financials rather than sport and I have never watched a football match, horse race or similar event. I am interested however in the term structure of the BetFair data which, to borrow an options analogy, exhibits a theta type property; though seemingly with a more linear structure through time.

... and I've never watched a financials race , so we're probably as far apart as you can get!
Also, I don't have the right psychological profile for betting in-running. I prefer to trade my market, then not return until it's all over to settle up one way or the other.

My question therefore is how does one begin to model event probability in this context. I know plenty of models for examining credit risk, market jumps and other binary events but don't know how one would apply these to sport. Would it be appropriate to do so? Is there a branch of study in this area and can you point me in the direction of any information, papers or other resources online?

What I can say is that IMO the 'organic' nature of football makes it very difficult to model. But of course that fact is true for anyone laying the odds too. All of my plays rely upon little more than mean averages . . . given circumstances x, the most likely outcome will be y.

Talking sports betting in general, being distanced from and having no emotional investment in the contests can be a big positive. With football in particular, many individual match odds can get skewed merely because of the weight (or lack) of support each team has. A situation which is excellent news for the 'averages' player.

I do enjoy a good football game, but horse racing does little for me. Today for example I have two racing trades in progress, but as to what horses/jockeys are running, or what wins which race, at what odds - I probably won't even bother to look. Unless I'm processing new data I rarely look further than the best quote to strike the bet . . . then what it makes-up to at the end of the day.

Only guessing, but if you gather enough data I would have thought something from your financials statistical toolbox would probably fit the bill?

 

[Data Miner]

I've a football interest this evening, trade #10 of April. So might be a suitable opportunity to briefly outline the football markets I go with;

Superiority:
Often not even labelled in the spread firm's quotes, but merely the number of goals superiority of one team over the other. These quotes are usually the very first one in a spread firm's listings.
e.g. PSV v Liverpool 0.1-0.3
tells you that this quote has PSV ending the game between 0.1 & 0.3 goals in front of Liverpool.
If the odds makers think the advantage is with the away team, the spread is not displayed with negative numbers, but the teams are reversed with an indication the 2nd team is the home side;
e.g. Liverpool v PSV (home) 0.0-0.2 (Quote suggests Liverpool to win by between 0 & 0.2 goals)

Total Goals:
Simply the total of goals scored in a game.

Mini-performances:
Only quoted for non-televised games. Points are awarded to each team on the following basis;
15 pts win
5 pts draw
0 pts lose
10 pts for each goal scored
5 pts if team keeps a clean sheet
MINUS 15 pts for each man sent off

example: PSV 2 Liverpool 3, PSV have a man sent off.
PSV mini-perf (2*10) - 15 make-up = 5
Liverpool M-P 15 + (3*10) make-up = 45

Oh yes, my speculation for this evening . . .

Stevenage v Kidderminster BOUGHT total goals @ 2.6 (SP) 25pts


Happy spreading

 

[Data Miner]

04 April 2007
+325.65 b/fwd

Today's racing;
11] SOLD Catterick SPs @ 45 (SP) 3pts

===== Previous trading results (from 03 April 2007) =====
+186.95 b/fwd

8] SOLD Wetherby SPs @ 53 (IG,SX) 3pts MU=37.1, +47.70
9] SOLD Southwell Favs @ 67 (SX) 3pts MU=40, +81.00
10] BOUGHT total goals Stevenage v Kidderminster @ 2.6 (SP) 25pts MU=3, +10.00

Day's return +138.70, +325.65 to c/fwd