• Hi Guest Just in case you were not aware I wanted to highlight that you can now get a 20% discount on Inform Racing.
    Simply enter the coupon code ukbettingform when subscribing here.
    We have a lot of members who are existing users of Inform Racing so help is always available if needed.
    Best Wishes
    AR
  • Hi Guest Just in case you were not aware I wanted to highlight that you can now get a free 7 day trial of Horseracebase here.
    We have a lot of members who are existing users of Horseracebase so help is always available if needed, as well as dedicated section of the fourm here.
    Best Wishes
    AR

Flatstats etc

Hi arkle55 arkle55

mlmrob mlmrob was on the VDW Forum (Same Username) run by Skyrack towards the end of our stay on that Forum and was posting the same kind of quality analysis as he does on here. Not really bothered about the cream from Flatstats having different usernames as cream always rises to the top, so enough said:handgestures-thumbup:
 
Mlmrob if you are looking in did you use the same user name on Ukbettingtips.

Arkle

Shhhhh, it's the place that must not be named.......... the betting equivalent of that famous Scottish play that supposedly you shouldn't say...........
 
Flatstats Members JTH, Dylan, Scipio & Wim

Dylan was the guy that first got me interested in using systems all those years ago. From what I remember most of his were for National Hunt - bit ironic as the site was called Flatstats :) The other names also sound familiar.
 
Another name was Many or Manychester maybe something similar. He used to put up lots of big priced winners, i think he used to cover national hunt flat races from a breeding angle.

Arkle
 
Another name was Many or Manychester maybe something similar. He used to put up lots of big priced winners, i think he used to cover national hunt flat races from a breeding angle.

Arkle

Manychester posts his selections on Twitter and like a few other regulars was posting less towards to end of my association with the Flatstats Forum. He tended to post in the last few minutes before the race as his main incentive was building up his Twitter Following. Manychester was good T weeding out the big price winners and as far a s I am aware was not a Paying Subscriber of Flatstats. like you say Arkle was very much into NH Racing and often picked the Irish Races
 
Database Optmising for Speed
One of the features of FlatStats is that we are very fast. Every report is run online so that it doesn't really matter what type of PC you have (unless you are using very old browsers) the stats reports will be generated in milliseconds (a few seconds for very big RSB reports).

We use specific techniques to optimise the server for speed. The first thing is to ensure the processor is as fast a possible (not necessarily zillions of cores but fast clock speed), the second thing is to use plenty of RAM (more RAM than the total GB of files on the disk! and run all the database reports in RAMDRIVES / Memory Engines), the third thing is to use SSD drives, and fourthly the O/S is highly tuned.

We use CentOS, MariaDB (a faster variant of MySQL) and Harbour (a modern day version of Clipper / dBase for those who know that stuff from the 80s and 90s). Basically nothing is Microsoft and nothing is bloated! Everything is run as a bare essential to ensure speed is as maximum as possible.

Pre Calculated Fields
The database is kept as small as possible with subfields used to reduce the size.

e.g. The FlaStats results database has hundreds of basic fields, but rather than have a column1, colum2 .... column700 ... we merge fields into one.

Take Trainer Record at the course. For this you can run a query for the trainer at the course and then count the number of wins and runs, calculate the strike rate, sum the profit, calculate the ROI etc. Now if hundreds of people are doing this then there are hundreds of these code snippets being run and hundreds of database I/O going on. It is far better to pre calculate 'M Johnston ROI% at the Course'.

So rather than having this

Course, Trainer, Wins, Runs, ..... (code to select these records and then code to calculate the fields on the fly) we do this

ST_COURSE = 27 Character Field, which contains the Trainers Record at the Course

For Mark Johnston at Hamilton his ST_COURSE field is

01090195047000081.120097.16

Now what the hell does that mean!

0109 = the number of wins at the course
0195 = the number of places at the course
0470 = the number of runners at the course
0008 = the ROI at the course
1.12 = the A/E at the course
0097.16 = the expected number of wins at the course

All this is pre calculated before the race. A techinque like this is a one shot update, which takes literally just a minute or two to go through hundreds of thousands of combinations of trainer and course.

The database engine which performs the pre calc is Harbour. It is very fast, and updates the final read only / RAMDRIVE MySQL tables.

All that is required client side is to now extract the sub sections and display it for the user.

e.g. to see the A/E for M Johnston at Hamilton we just use a substring command either in MySQL or PHP.

Of course if you are doing clickbox type searches then this is different - you need dynamically calculated reports, which can not use pre calculated fields. You therefore need to ensure that the MySQL is tuned to the max.

1. Don't use InnoDB for your front end stuff. It may be good (safer) for transaction logging but you just can't beat MyISAM run from Engine = Memory
2. Make sure you have indexes only for main fields (name fields). There is a tendancy to index everything but this can be counterproductive.
3. Add more RAM. Can't point this out enough. If you are running your DB on a laptop or desktop then the O/S is going to take a chunk, you need a chunk for file caching too, so all three will be fighting for RAM. Add as much as your device can possibly take.
 
Last edited:
Thanks a lot for that post, baycolt baycolt. Fascinating stuff! The multi-character column is something I've been mulling over for my own French DB. I'm using Excel as my front end and it's fast at getting the data in from SQL Server for further analysis. It would be very inefficient for multiple users of course, but I can live with the time overhead in order to be able to format the output exactly as I want.

Your 27-character field brought me back to the early '80s when I first started putting racing data into a "PC". I use the inverted commas because it was a Sinclair ZX-81 with a membrane keyboard and its pushing-the-boundaries 15 K RAM add-on. The only way I could squeeze data in was by bit-manipulation in the different registers. All programming was written out longhand in machine code to get the logic right, translated into binary and then input to the ZX-81 in long comma-separated strings of the binary data. It was extremely laborious, but I was fascinated by what I could make the little computer do for me. I think I managed to get in two years of data, culled from the yearly Form Book, for Guy Harwood, Henry Cecil, Dick Hern, Michael Stoute and Peter Walwyn. Each year of data for each trainer was stored on a cassette tape and then reloaded into the very basic analysis module. It was great fun until life intervened and a 25-year hiatus from horse racing ensued. Got back to it again in 2010 and I'm still in awe at what we can make these machines do for us.

Ray
 
Ray, if someone does put together a very good French database (if that is the right term to use) it will have a value as the French as far as i am aware the French do not have anything like a R/Post paper. There is as you will know Paris Turf and i think one other paper, the name escapes me. Paris Turf, the last time i looked is very basic form, much worse then a R/Post, probably because the French prefer trotting. I wonder if Timeform cover French racing.

Arkle
 
Hi Paul,

Yes, it's true. Sixty percent of all racing in France is dedicated to trotting. It's not something that interests me and my limited understanding of it is that it's very strategic in how a race is ridden. Calimero1250 Calimero1250 knows a lot more about it than I and can add more details. There are several papers, apart from the P-T, that give good coverage of racing: L'Equipe (perhaps the one you were thinking of?); Bilto; Tierce Magazine and Geny Courses are the ones that come immediately to mind. However, there seems to be an emphasis on the exotic bets over here, particularly the Quinte+ where you name the first five home, usually in a 14-to-20 runner handicap. It's difficult enough to come up with the first home, much less the first five home, so it's not something that interests me either. That said, a measure of the popularity of the exotic bets is perhaps the PMU takeout from these bets. The Tierce (first three home) 37%, Quarte (4) 35%, Quinte+ 33%, Couple (2) 27% and the single win bet (Simple) a mere 16.3%. Quite a difference. In that context, I'm not so sure that a French database will have much value. Exotic bets forecasting doesn't really benefit from the exploration of a database. The PMU takes about 90% of all bets, but it's not often that the Simple pool on the PMU exceeds 500,000 euro (and that includes all bets on-course), whereas the exotic bets will have pools of millions.

Ray
 
Ray, i think Phil Bull was the first to offer commercial ratings/speed figures (Chesham will know a lot more than me on this) and then Timform was born i think. If you can offer / be a little different than say Timeform as a example that way you could be unique in what you offer. You would have a platform here and other sites to get feedback on what people would like to use as a tool and as things improve and you have enough paying members you keep improving the tools your membership will keep growing. If the French betting market is not being tapped into you could be on to a winner. Just like here from what i can see French Racing is also very backwards at coming forwards and just like here it is purposely kept that way so exploitations of the racing industry can take place. It could be a money spinner at the right price.

Arkle
 
Twenty years ago, I would have gladly run with this business angle. But I'm a retired man of leisure now and life in the slow lane is too appealing to throw away on setting up and running a business! I'd have to go from the freedom of 24 hours a day belonging to me to spending all day dealing with customers who can never be satisfied. No thanks.

I'm a data junkie, though, and the fruits of my labours will be made available here - for free - if anyone thinks it will be worth using.

Ray
 
Twenty years ago, I would have gladly run with this business angle. But I'm a retired man of leisure now and life in the slow lane is too appealing to throw away on setting up and running a business! I'd have to go from the freedom of 24 hours a day belonging to me to spending all day dealing with customers who can never be satisfied. No thanks.

I'm a data junkie, though, and the fruits of my labours will be made available here - for free - if anyone thinks it will be worth using.

Ray

i for one will soak up your knowledge like a sponge rpjd99 rpjd99. Data systems are a totally new thing for me, but amazed at how my labourosly slow brain is enjoying the process.
 
Hi turfburger turfburger,

You're very welcome to any help I can give you along the way. However, as this is a FlatStats thread, common courtesy demands that I stop writing about myself and my daily data fixes. If you do have any questions, you could start a new thread and we could deal with them there.

Ray
 
... back to the early '80s when I first started putting racing data into a "PC". All programming was written out longhand in machine code to get the logic right, translated into binary and then input to the ZX-81 in long comma-separated strings of the binary data. It was extremely laborious, but I was fascinated by what I could make the little computer do for me. I think I managed to get in two years of data, culled from the yearly Form Book, for Guy Harwood, Henry Cecil, Dick Hern, Michael Stoute and Peter Walwyn.
Ray

I took the Commodore route! Started with PET and then the C64 with standard tape drive and later a 5 1/4 floppy drive.

The upgrade from tape to the floppy drive was like going from a penny farthing to a model-T ford. The speed, random access and capacity were outstanding compared to tape. You didn't have to listen to the screeches either.

I remember that C64 floppy drive had a flaw (or more of an easter egg) that allowed to you double the usage of a standard disk. The drive was single sided, but you could turn the floppy upside down, cut out the read / write notch on the wrong side and thus use the disk upside down thus doubling the capacity from 170KB to 340KB. I loved stuff like that and got tinkering with the I/O ports and the insides and such.

The first thing I did was to create a tape duplicator. Yes, it was possible to duplicate games on a dual cassette deck but some did not work and were rumoured to have some kind of protection to prevent it (a continous frequency, which the dual cassettes did not like). What I notched up was an adaptor to fit between the tape drive plug and connector. It was a more direct feed off the the tape deck that allowed a cleaner signal for the recording deck. It was then easy to duplicate games at the time.

Another device I created for the C64 was 'The Red Button Reset Switch". This was a device that allowed you to warm boot the C64. It consisted of a 6-pin din plug, a resistor and a small single pole press button switch, which just so happened to have a red button. When the button was pressed the Reset pin was grounded thus invoking a warm boot. Now why would you want to warm boot via a button on the back of the device? Because some games disabled the keyboard warm boot.

The games designers did not want users modifying the code so this was a challenge that had to be taken up. The reset device was built and dozens of games were hacked in order to over ride game limits. I spent hours hacking away at the machine code to find the right memory locations for things like game level indicators and ship lives. From that I worked out the poke codes to allow users to instantly start on level 5, or to have 255 lives in space invaders. That red button reset switch with a printout of the poke codes to change the game settings sold like hot cakes.

I made a few quid from those and that allowed me to go down the more traditional "IBM" PC route with proper hard drives (20MB). From there I got into proper dBase databases. I set up a nice line in programming supplying ready programmed PCs to Video Rental Stores. I did freelance DB work for other industries too. All this allowed me to purchase proper hardware and invest more time in building up DBs. So you could say it all started with a small red button reset switch.
 
Hi Chesham, nice post. I can not get my head around pos vs pos in market. Does this mean, horse last time out should have been anything from 1st to 3rd fav and its lto finishing postion should be no worse than 5th. Example, if horse was fav lto then its finishing position should be no worse than 3rd or if horse was sent of 3rd fav then its finishing position should no worse than 5th. Im not to sure if i am understanding what is being said correctly. HRB says '' a score based on how the horse has finished in it's career compared to how it should have done based on it's starting price.''

Thank you
Arkle
 
Hi Chesham, nice post. I can not get my head around pos vs pos in market. Does this mean, horse last time out should have been anything from 1st to 3rd fav and its lto finishing postion should be no worse than 5th. Example, if horse was fav lto then its finishing position should be no worse than 3rd or if horse was sent of 3rd fav then its finishing position should no worse than 5th. Im not to sure if i am understanding what is being said correctly. HRB says '' a score based on how the horse has finished in it's career compared to how it should have done based on it's starting price.''

Thank you
Arkle

Hi Arkle

It is a a calculation in the HRB RSB that Rates the career performance based on SP and Finishing Positions. In this case it is only based on the horses first run. I don't know the formula for the Rating though.
 
Hi arkle55 arkle55
with reference to PosVPos in market, I have done some experimenting, with the following 'system'
upload_2014-9-4_21-42-49.png
You can check all of the qualifiers and you will see that all selections LTO finished the race in the same position as their ranking in the market ie favs won, 2nd favs came 2nd etc.
A value 0f -1 means that the horse finishes 1 place lower than its market position, so if it was fav it would finish 2nd, conversely a value 0f 1 would mean that the horse would finish one place better than its market rank.
However, if a horse has had more than 1 previous run I think the value is worked out as an average by adding together the value for each race and dividing by the number of races.
 
Back
Top