Trading What I see !?

[ShaunWhite]

I think I’m correct in saying that ShaunWhite uses his own custom built software now too, and if you look at the inspirational graph that he posted on the credibility thread, his profits have reversed sharply and have been heading to the moon ever since.

I didn't show the Y-axis scale. It's more like heading for the loft than the moon.

Software alone won't make a profit and people do make good money using Guardian. Guardian is really good and getting better.

The reason I have my own stuff is primarily for the data collection, analysis and testing against the saved api. And in order to reduce the risk moving from test to live, I just switch the data source to live (from saved) and the orders to live (instead of simulated). I could just recode it in Guardian but I do lots of logging too and my results analysis plugs into that.

Basically it's not Guardian that's the isuue, as I said it's really good and dead easy to knock things up quickly. But running a strategy is only one step in the process and I felt I needed to build a decent testing system and decent results analysis around it. The fact it can also execute bets is almost secondary.

I'd say initially don't build a Guardian, build something to test strategies quickly and most importantly accurately then going live holds no surprises. BA is a great transaction processing engine but obviously can't also be a great pre-trade and post-trade system too, because there's too many ways to skin those cats. My background is 25yrs in City IT, all I'm doing is bringing some of that methodology over to my sports betting. What I had didn't match what I wanted so rather than just making do I dealt with it.

I really don't want to be held up as an example, the wheels might fly off tomorrow and I'm doing nothing that lots of other people aren't doing and I've had a lot of guidance and help from some really talented guys. Without them I'd be nowhere.

So, typically long comment but I hope that puts that topic to bed because I'm bored with retelling my 'story' now and I'm certain most other people are.

[gazuty]

... people do make good money using Guardian. Guardian is really good and getting better.

+1

(Post not directed to Shaun but to those who say guardian is not sufficient or that successful traders build their own software. I can only give my own anecdotal experience in response).

When I started using BetAngel, Guardian had some functionality that I needed alongside my manual efforts (Guardian provided me with my five seconds to go failsafe). I then spent time with Excel to try and get to full automation across many markets. There were some limitations in Excel (or in my own abilities with Excel) that meant I couldn't achieve what I wanted to achieve in Excel.

Then, through the continual development made by Peter and the BetAngel team, the day came along so that what I had been trying to do in Excel was now achievable through Guardian. That enabled me to go from a person who dabbled manually on Saturdays and Sundays in front of a screen (and annoyed my wife becasue I was tied to a computer) to a person who could load up every single market into Guardian and apply an automation that worked on Australian racing. Pretty quickly after that I also developed something that worked on Australian and UK greyhounds. Then I moved from relying on my own computer at home to a VPS set up so that I could simply log in, load up markets into Guardian from anywhere and click the "apply rules to all markets" button. Nirvana.

Obviously different markets have different characteristics and so I then focused attention on US horse racing markets and UK horse racing markets. I eventually developed an automation that works in guardian for UK horseracing markets and while I made some progress on US horse racing I gave up because UK horse racing gave me all the profits I needed.

So much development has been put in to Guardian over the years that it is now way more sophisticated and can execute things based on variables that I don’t ever consider.

Over time I've looked at other markets (politics, basketball, football, tennis etc) but not on the same systematic basis as horse racing and greyhounds. I'll get to those other markets in a serious way one day.

[Emmson]

Good to know above as my self-schooling on Excel is proceeding very slowly, only at nested If functions - miles away from Macros & VBA

[Anbell]

Good to know above as my self-schooling on Excel is proceeding very slowly, only at nested If functions - miles away from Macros & VBA

Shaun had some good advice the other day: Use AND and OR, rather than nested IFs, where possible.

[Emmson]

Good to know above as my self-schooling on Excel is proceeding very slowly, only at nested If functions - miles away from Macros & VBA

Shaun had some good advice the other day: Use AND and OR, rather than nested IFs, where possible.

Yeah I saw that, said they are easier to read. Made a mental note of that but now will write it down.

[Derek27]

Good to know above as my self-schooling on Excel is proceeding very slowly, only at nested If functions - miles away from Macros & VBA

Shaun had some good advice the other day: Use AND and OR, rather than nested IFs, where possible.

Yeah I saw that, said they are easier to read. Made a mental note of that but now will write it down.

I've never found either of them to read, but VBA is once you've learnt it.

[ShaunWhite]

Yeah I saw that, said they are easier to read. Made a mental note of that but now will write it down.

It's just more how we think anyway....
IF this AND that then do whatever, otherwise don't.
A bit more natural than
IF this then IF that do whatever otherwise don't, otherwise don't

Simpler still would be to make a cell contain true or false depending on your condition. Then say IF cell = true then...... Makes debugging easier sometimes if you don't cram so much into one formula. Lots of ways to skin a cat though. Ain't gotta be pretty if it works.

[goat68]

So this week a mere -£57 for bot, rubbish! Convinced it's a backtest fitted piece of rubbish, and is going in the garbage...

So this bot was my attempt after many iterations of value trading based on my automated view of manual trading, rather than the more traditional "quant" high frequency auto trading value bot.
My quandary is i'm not sure it's vaguely possible to create a bot to replicate order flow manual trading, what with all the soft skills the brain goes through when manual trading. As opposed to "quant" value trading, in which I really can't see how you determine what is value on the "macro" scale. How can you really identify 2-3ticks of value at any point in time? the market noise is larger than that!

Doing my head in. People say it's good this trading is hard, otherwise everyone would be able to do it, so just keep going longer than the others who give up, but how much money must I lose to get there? and will I get there? if i'm not going to get there, then it's a waste of time and money!

[Kai]

So this week a mere -£57 for bot, rubbish! Convinced it's a backtest fitted piece of rubbish, and is going in the garbage...

It appears you have downloaded the wrong piece of automation Mr Goat! Try downloading one from the "Profitable Automation Forum Section"

[goat68]

Part of the problem, and I don't mean this in a nasty way, is this forum is full of mostly "manual" traders, and Peter himself is mainly manual, so a lot of the advice and videos is manual based. The reality I think is it is very very difficult to fully automate manual strategies.
The other reality is automated successful traders obviously keep their secrets close to their chest!

[goat68]

"toys" not in "pram"

[Kai]

Sorry to hear that Mr Goat, only thought you might need cheering up!

That is a good point about manual traders. However, I am of the opinion that even auto traders often overshare. Problem is, the useful bits and bobs that get shared probably get lost in those never-ending Today's threads.

[jamesg46]

Part of the problem, and I don't mean this in a nasty way, is this forum is full of mostly "manual" traders, and Peter himself is mainly manual, so a lot of the advice and videos is manual based. The reality I think is it is very very difficult to fully automate manual strategies.
The other reality is automated successful traders obviously keep their secrets close to their chest!

You can back test all you want but it's useless, unless you're very specific with the variables which by all accounts are different by the milliseconds.

Manual is easier because it deals in reality, which is where you started (name of thread).

Believeing what you see is sometimes mysterious.

[ShaunWhite]

As opposed to "quant" value trading, in which I really can't see how you determine what is value on the "macro" scale.

Quant is just anything algorithmic, not always just value seeking. So any botter/automation guy is a 'quant' and if nothing else it's a damn sight easier linguistically.

but how much money must I lose to get there? and will I get there? if i'm not going to get there, then it's a waste of time and money!

If you're not getting any pleasure out of it then it's just a straight value judgement, how do you rate your chances? "If I lose it's a waste of money" makes me think you're not a natural gambler If you feel like quitting then fair enough, but switching to £200 stakes didn't help, prior to that you'd spent about £2 a day. Don't get too stressed you've only tried one idea, it's a familiar start.

Making a bot that emulates the manual style of continually assessing your position was a bold start. Especially if you can't do it manually in the first place. As is predicting the price 3 minutes away when guessing 10s is hard enough. How about just trying to steal a tick or two around crossover points, a BounceBot. Or maybe a SpikeBot that leaves offers out of range in every market 247 waiting for spikes (use auto for what auto does best). You need to watch a few hundered markets and look for those little things. Rates of change eg accelerating or decelerating steams/drifts might be interesting. I did a calculus course on Udemy because I wanted to look at that and never did. Well I say 'did', I quit because I thought I wouldn't need the primer and it was all too heavy. Quant trading isn't very exciting really like clicking cash money around is, it's a bit schooly.

[wearthefoxhat]

Part of the problem, and I don't mean this in a nasty way, is this forum is full of mostly "manual" traders, and Peter himself is mainly manual, so a lot of the advice and videos is manual based. The reality I think is it is very very difficult to fully automate manual strategies.
The other reality is automated successful traders obviously keep their secrets close to their chest!

Worth repeating, for me, semi-automation is preferred.

The ideal scenario is to find an entry point that your automation can trigger, more than once, if the set up requires it, then manage the trade manually. This could include, breaking the exit trade down in blocks, or exit in one ago. Maybe the data you've gathered, has something that supports that.

Over time, i've found the best trading opportunities are in-play. Pre-trading on horses, is pretty much carved up by better traders and sharper bot programmes, so IP, which is a growing market, is where the action is. Many avoid it because of it's fast moving, eye popping swings.

The prep for IP, centres more around how you would visualise a race and then how it starts to pan out. Also, identifying selections that are fancied in the market, but in effect no hopers under current conditions, present strong L2B opportunites too, and let it ride on occasions. Again, this suits me personally backed up with my many years of watching and studying the horses. (wasted youth?)

The sub-forum within Bet Angel is a good starting point.

viewforum.php?f=52