Trading What I see !?

[goat68]

So for example "Lay the poor starter bot": viewtopic.php?f=52&t=12100
What I think about with this, is so the theory is horse drifts 10ticks after start, but to make money from this the price still has to be value? regardless of the fact it's slipped 10ticks. How do you really know? don't think you do? In reality it's slipped 10ticks due to a poor start, and the price then being offered is based on Exchange backers & layers opinions... If everyone is actually using this bot, i'd actually do the opposite...?

[wearthefoxhat]

So for example "Lay the poor starter bot": viewtopic.php?f=52&t=12100
What I think about with this, is so the theory is horse drifts 10ticks after start, but to make money from this the price still has to be value? regardless of the fact it's slipped 10ticks. How do you really know? don't think you do? In reality it's slipped 10ticks due to a poor start, and the price then being offered is based on Exchange backers & layers opinions... If everyone is actually using this bot, i'd actually do the opposite...?

The principle of using this type of bot and laying a "value" selection, will depend on it having been overbet pre-off. ie: The sentiment is that a runner will establish a quick lead and DOB/HOB/QOB.

The intial research could be to look at the pace cards and if there was a steamer pre-off on a runner that is likely to lead, and then it starts poorly, you have a likely candidate. If it starts well, then the bot will not fire as the selection's price will drop further.

[foxwood]

So for example "Lay the poor starter bot": viewtopic.php?f=52&t=12100
What I think about with this, is so the theory is horse drifts 10ticks after start, but to make money from this the price still has to be value? regardless of the fact it's slipped 10ticks. How do you really know? don't think you do? In reality it's slipped 10ticks due to a poor start, and the price then being offered is based on Exchange backers & layers opinions... If everyone is actually using this bot, i'd actually do the opposite...?

That bot has nothing to do with "value" imho. It's simply trying to make a trade out of a commonly observed market reaction to a real-world event - ie when certain runners are slow out of the gate they will often drift quite a bit higher. It's not an outright lay which would be a "value" decision.

I haven't downloaded it but the specification is quite clear in the first 3 paras - a bit that may be missing is to take a smaller green or to red-up when the runner fails to drift enough to make the trade. That would be down to the user to fine tune as with all Dallas' offerings.

You would need to be selective on race type, distance, timing, etc and tune the mechanism in practice mode. Then, when that looks like it is working correctly, go live with small stakes. Rinse and repeat based on that experience.

[decomez6]

My current bots are simple, i'm thinking i've gone too simple:
- WoM > 75% or <25%
- Price between 2 and 3.75
- TotalVolume 75% of expected start volume
Trigger: Back or Lay

+1
too simple ....but...on the plus side , easy to diagnose.
i personally would have such a set up for a servant when manually trading. it will monitor the price point pressure as i keep an eye on those spoofers in conjunction with the volume bar which is a better indicator of real volume matched.

i also see some problems with the boxes you tick before executing . they seem to be based on a false premise :

1. WOM
-how much of it is down to cross matching and spoofs ?
why 75% any historical evidence ? in the same race/event/going/jockey/owner/ratings/...............ETC any fundamentals could make or break it.
do you know why 75% is performing better in one race and not the other?

2. expected Total volume 75 %
how do you calculate this?
-an accumilatve average of the same race over how long of a period ?
do you consider seasonality and other conflicting fixtures that could dry up the volume flow E.g when another race is delayed and the money arrives late or do not arrive at all?

3. price 2 - 3.75
how do you choose when more than one selection have identical WOM , VOL and PRC...... you need more variables to determine a lay and rule a back vise/vasa

My take Novice take:
testing is great ! you are in the right place.......if i had your skills i would spend less time thinking about the testing and concentrate on putting some 'MEAT' on the 'Bone'
make the bot do some machine learning

have some good old fun and try step away from the blue screens for some time . things seem to work better when you are in a less tense state and you will look at the bot differently .

[arch4672]

My automation does well and its a simple but affective bot they are out there goat just need to keep things simple

My current bots are simple, i'm thinking i've gone too simple:
- WoM > 75% or <25%
- Price between 2 and 3.75
- TotalVolume 75% of expected start volume
Trigger: Back or Lay

I'm right pissed off, today bots have lost me -£95, 14 losing markets out of 20

If you are not enjoying it anymore stop, don’t waste any more money.

+1

Stop wasting your money and do something more worthwhile.

[wearthefoxhat]

So for example "Lay the poor starter bot": viewtopic.php?f=52&t=12100
What I think about with this, is so the theory is horse drifts 10ticks after start, but to make money from this the price still has to be value? regardless of the fact it's slipped 10ticks. How do you really know? don't think you do? In reality it's slipped 10ticks due to a poor start, and the price then being offered is based on Exchange backers & layers opinions... If everyone is actually using this bot, i'd actually do the opposite...?

The principle of using this type of bot and laying a "value" selection, will depend on it having been overbet pre-off. ie: The sentiment is that a runner will establish a quick lead and DOB/HOB/QOB.

The intial research could be to look at the pace cards and if there was a steamer pre-off on a runner that is likely to lead, and then it starts poorly, you have a likely candidate. If it starts well, then the bot will not fire as the selection's price will drop further.

Here's a quick example of what I mean.

As I write, the race has just ended.

Anemoilog.png

The only thing I tweaked on the original file was to nominate a selection and use only one selection to lay per qualifying race. In the first 2 races there were no matches as they started the race as expected.

anemoi.png

The key is to find selections that are fairly fancied where you can estimate their running style and pre-market moves. The bot does the rest.

[goat68]

Being studying various inplay data and strategies, but it all seems like I am guessing parameters and methods again... Which really means, i'm only going to find anything profitable by "luck", rather than knowing/understanding why something might be "profitable"/"edge".
I'm trying all sorts:
- short term triggers on all runners at prices above/below current inplay price
- using race distance to duration estimation to calculate the approximate time of the last 15seconds or so of the race, and trying to catch value in reversals and various other variations!
- using race competitiveness evaluation based on closeness of the top 4 BSP prices

I have NO bots running live now, and don't intend to until I am more confident

[goat68]

one observation, something might work really well one or two weeks, then be really duff the next 2 weeks!

[goat68]

Thinking about it, inplay it's mostly random price jumps based on what's happening, so to find an "edge" you'd need some "extra" info that not everyone else has eg:
- GPS location and low bet latency
- "fundamental" analysis pre-race which might give you an edge on how the race might run
I think trying to find something based "purely" on price action is near nigh impossible.
So I am going to try some "web scraping" maybe!

[ShaunWhite]

Did you look at TPD?

Stay positive, you've only been looking at in-running for about a week so it's not surprising you haven't had any breakthroughs yet.

[goat68]

Did you look at TPD?

Stay positive, you've only been looking at in-running for about a week so it's not surprising you haven't had any breakthroughs yet.

is TPD == GPS data?

[napshnap]

Goat, keep in mind that some venues have slower video (ATR covered) than others and it can influence price action.

[goat68]

Goat, keep in mind that some venues have slower video (ATR covered) than others and it can influence price action.

that's an interesting point, thanks

[goat68]

Did you look at TPD?

Stay positive, you've only been looking at in-running for about a week so it's not surprising you haven't had any breakthroughs yet.

google never fails! : https://www.totalperformancedata.com/

[ShaunWhite]

I have NO bots running live now, and don't intend to until I am more confident

First job is understanding how/why you misinterpreted the other backtests, then you can start to have confidence in them. It would be demoralising to make the same mistakes again.