price-prediction-bot

price-prediction-bot copied to clipboard

Hi, you can never backtest with the final odds as you can bet only till like 7-8 seconds before closure and a lot of pending transactions will be after your bet and this will sometimes make odds change from 2.5 to 1.3 and reversed, I backtested with the real odds and had a strategy that would bring me a lot of money, but when i tested it i never had the correct odds as when i betted in the last 5 secs the transaction would fail with 'betting too early/too late'.

Just to let you know. I have also written a prediction app in node.js and used some strategies, but never use backtesting untill you find a way to overcome the 5 seconds too late issue...

Hi, you can never backtest with the final odds as you can bet only till like 7-8 seconds before closure and a lot of pending transactions will be after your bet and this will sometimes make odds change from 2.5 to 1.3 and reversed, I backtested with the real odds and had a strategy that would bring me a lot of money, but when i tested it i never had the correct odds as when i betted in the last 5 secs the transaction would fail with 'betting too early/too late'.

Just to let you know. I have also written a prediction app in node.js and used some strategies, but never use backtesting until you find a way to overcome the 5 seconds too late issue...

I also faced a similar issue with my strategy. I tried everything from betting late to betting both sides then canceling one but nothing worked. I've been examining ways to scan the mempool for pending txs and then using that to estimate the real final payout and was making progress but had to stop to focus on other more pressing work. I have limited knowledge of how web3 and dapps work and would like to work on this with someone more experienced. If you're interested, mail me at: [email protected]

listening for pending transactions does get you a better look of the actual bets but still not 100% accurate. still, it gives much more up to date values than checking the contract.

listening for pending transactions does get you a better look of the actual bets but still not 100% accurate. still, it gives much more up to date values than checking the contract.

It does?! I thought as much that it would. How did you manage to get a better estimate of the odds from all that data, must have been quite messy. How accurate do you think it is, perhaps you could measure it?

Yes, pending transactions, but how do you overcome the ‘limit’ on the providers, you should be able to filter the contract pending transactions first because otherwise you get limited if you don’t have your own node server…

Verzonden vanuit Mailhttps://go.microsoft.com/fwlink/?LinkId=550986 voor Windows

Van: @.> Verzonden: 23 February 2022 13:04 Aan: @.> CC: @.>; @.> Onderwerp: Re: [jeffersonlicet/price-prediction-bot] Backtesting will never give real results (Issue #6)

listening for pending transactions does get you a better look of the actual bets but still not 100% accurate. still, it gives much more up to date values than checking the contract.

It does?! I thought as much that it would. How did you manage to get a better estimate of the odds from all that data, must have been quite messy. How accurate do you think it is, perhaps you could measure it?

— Reply to this email directly, view it on GitHubhttps://github.com/jeffersonlicet/price-prediction-bot/issues/6#issuecomment-1048713057, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ACOZWZ6ZTL57S2GOWNN6M23U4TENBANCNFSM5M6WBEUQ. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub. You are receiving this because you authored the thread.Message ID: @.***>

in pending transactions i filter for all betBull and betBear calls and those methods also have the amounts given. therefore, whenever the events are triggered i just add the values to the ones i have and i compute the multiplier. this way i see all the bets sent in, up to the last possible block. the thing is that not all of those transactions will actually make it in, therefore i don't have the 100% accurate picture of the final outcome

i need to mention that i do run my own node

do you? what does it cost to set it up? And do you get all the pending transactions, then, i heard you need to have full node servers all around the world to do that?

i don't know the technical aspects behind setting it up as a friend of mine did all that, sorry. we rented a server located in the US

do you have discord so we can chat about that subject more, I am interested in the topic? I am personally kind of good with the web3 thingies, perhaps we can brainstorm on the subject...

no, i don't. i have skype :) i am old fashioned. send me an email and i'll give you my username

it's not about being good in web3 thingies - which i also am :) - but about coming up with a good strategy

I just sent email to ish*@gmail, is that your email?

nope. click on my username, get to my profile, it's written on my page

in pending transactions i filter for all betBull and betBear calls and those methods also have the amounts given. therefore, whenever the events are triggered i just add the values to the ones i have and i compute the multiplier. this way i see all the bets sent in, up to the last possible block. the thing is that not all of those transactions will actually make it in, therefore i don't have the 100% accurate picture of the final outcome

In my experience, anything placed within the last 10s fails.

Last time bets will fail due to block timing, great to see this discussion here, I did analytics over the backtested data, seems 50/50 in general :)