Let's discuss strategies to improve success rate.

[tankiatwee]

I finally got the bot running and am happy with the way it works...but not the success rate. I set Threshold at 80 and still success rate goes below 50%.

Since martingale is not involved, I like to find out if fellow users have success using the bot in different ways.

For example, I could set reverse betting, or increase bet size together with decreasing daily profit goal so I could be done for the day quickly before overall result gets worse.

Has anyone found out if a particular time period of the day (from what time to what time) yields high success rate?

Any other recommendations are welcome. Please share, thanks.

[rootisareservedword]

I am currently testing multiple signal strategies in simulation mode, including combinations of different time frames other than the default. The default is to analyze 1 minute and 5 minute technical summaries from TradingView. I am testing 1M/5M, 1M/5M/15M, 5M/15M, 15M/30M, 1M/15M and 1M. Minimum threshold set to 90; reverse and non-reverse betting for all timeframes.

I feel like I need a minimum of 1000 trades per strategy before I can even begin to determine which one has the highest success rate, but am nowhere near that. Win rate in my opinion is not the most important factor, but rather risk/money management. You can still be profitable with a win rate of less than 50% if you have a sound money management strategy.

This bot is great as is with many options for customization like daily goal TP/SL, reverse betting, etc. But I think it could have great potential with even more options in the global config file. Some ideas:

• Martingale
• Anti-martingale
• Fixed ratio per trade (ie risk 1% of account per trade)
• Set active trading hours
• Increase/decrease bet size based on X number of consecutive wins/losses
• Hybrid mode. Run in simulation mode, but if bot loses more than X number of trades in a row, switch to non-simulation mode (or do the opposite)

It would also be nice if the bot could write and append results from each run to a CSV file. This would allow more efficient means to analyzing strategies. You could see total number of trades, wins, losses, consecutive wins/losses, and possibly find a statistical edge just by analyzing results.

I will share my results with strategies once I hit 1000 trades per strategy.

Off topic. In addition to QuickNode which offers 10,000,000 requests per month for the free plan, I also found NodeReal which offers 350,000,000 monthly requests for free. This could help when testing multiple strategies.

[tankiatwee]

All these are great suggestions if not for my main concern that lmusarella may possibly have taken his programming expertise as far as he can with this bot.

If another programmer steps in, it may or may not take more time for them to familiarize the existing code.

Anyways, we try to pitch in contributions.

[lmusarella]

Hello, yes unfortunately the success rate is not optimal, compared to what other BOTs would like you to believe. My goal was to create something stable and secure that I could use. I'll try to integrate the Martingale strategy and CSV management as soon as I have some free time. Thank you so much for the comments and above all for the suggestions you are giving me, I will certainly take them into consideration.

[tankiatwee]

Please download the Word file and Excel sheet for some great Martingale/non-Martingale ideas.

Martingale.docx
Only bet 70 percent.xlsx

[lmusarella]

Hi all, I just updated the code adding to the configuration file the possibility to use a martingale or anti-martingale in a quite simple way.

[spacechicken]

Thanks very much I will load it up today - legend!

[lloydwoods]

Loving all this activity! @lmusarella I still will throw a tip your way, someone hacked my PC the night we last spoke and drained by BSC wallet (it was my fault.....), but I will not forget you. I have to get some more funds into the BSC chain before I play again, but am very excited for the new strategy option. @rootisareservedword please let us know if you find a great strategy, there has to be a way to game this system to make some extra coin.

[lmusarella]

The question is legitimate, I made sure that if you operate with the martingale or the anti-martinagala, the BOT operates in a linear manner without having rounds waiting. So if I bet on one round, I don't bet on another round until this one is over. Il gio 27 apr 2023, 04:46 mitchbarks ***@***.***> ha scritto:

…

I plan to use this bot with the higher quote strategy and martingale, though I did have a question about how the bot knows it has lost or won the previous round when martingale is being used. You have to make a bet on the next round before the current round has finished, so how can martingale be used properly? The only way this can work is if the bot bets on only odd or even rounds, so essentially skipping a round so the martingale strategy knows if it has won or lost its previous round it bet on. — Reply to this email directly, view it on GitHub <#13 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIRDYVDKHNTNJSQ3S7T2BFDXDHMZTANCNFSM6AAAAAAXEBNMAQ> . You are receiving this because you were mentioned.Message ID: <lmusarella/Pancakeswap-Prediction-Bot/repo-discussions/13/comments/5737897 @github.com>

[mitchbarks]

Awesome~ thank you very much for this! The bot looks great, really appreciate your work on this! I will donate some money when I've got some BNB from the bot! Love your work, keep it up my friend! Mitch.

…

On Thu, 27 Apr 2023, 4:37 pm Luca Musarella, ***@***.***> wrote: The question is legitimate, I made sure that if you operate with the martingale or the anti-martinagala, the BOT operates in a linear manner without having rounds waiting. So if I bet on one round, I don't bet on another round until this one is over. Il gio 27 apr 2023, 04:46 mitchbarks ***@***.***> ha scritto: > I plan to use this bot with the higher quote strategy and martingale, > though I did have a question about how the bot knows it has lost or won the > previous round when martingale is being used. > > You have to make a bet on the next round before the current round has > finished, so how can martingale be used properly? > > The only way this can work is if the bot bets on only odd or even rounds, > so essentially skipping a round so the martingale strategy knows if it has > won or lost its previous round it bet on. > > — > Reply to this email directly, view it on GitHub > < #13 (comment) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AIRDYVDKHNTNJSQ3S7T2BFDXDHMZTANCNFSM6AAAAAAXEBNMAQ > > . > You are receiving this because you were mentioned.Message ID: > <lmusarella/Pancakeswap-Prediction-Bot/repo-discussions/13/comments/5737897 > @github.com> > — Reply to this email directly, view it on GitHub <#13 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A27PITBMDP6QE7HGZBWJBI3XDIH2TANCNFSM6AAAAAAXEBNMAQ> . You are receiving this because you commented.Message ID: <lmusarella/Pancakeswap-Prediction-Bot/repo-discussions/13/comments/5739148 @github.com>

[lmusarella]

Thanks, I'm glad you like it.

obviously if you like the project and want to support it, just one star here on github is enough.

Anyway this is my wallet: 0x0C40e4F3606aE2F41DAc9bB67606f41284755350 :D

[tankiatwee]

Today is the first full-day demo run of PCS Prediction with martingale.

Base bet: $4, total gain: $53, W/L = 32/25, 56% win rate

Some things I observed:

1. I always suspect Prediction's 5-minute blocks runs slower than your standard time, so you will not expect the bot to execute 12 trades every hour.

2. @mitchbarks says it right: the bot trades alternate rounds in order for martingale to work. I found it a good thing as it avoids the possibility of consecutive 5-min block losses...unless the gods really hate odd or even numbers for the day.

3. Speaking of odd/even numbers, if the bot skips a trade because of low Threshold, it takes the next immediate trade and thus 'breaks' the alternating pattern.

4. The demo had a moment of going to step 5 in martingale (just this once). Starting capital has room for 9 doubling-ups. Good or bad?

5. I just don't think above-50% win rate is always reliable everyday, but I have to demo further. Any other suggestions to increase $$$ despite low win rate are welcome, like is there a "sweet spot" number for Threshold?

6. I need to mention the bot has halted twice without warning, so I have to restart it. Let's say it's round #166765 on the website but the bot is stuck at #166760.

Overall, thanks @Imusarella for a great bot!

[spacechicken]

Mine died on 166760 as well lol - it may have been Pancake swap

[lmusarella]

If you like the idea, I inform you that I'm developing an application that you can launch in a similar way, but that will have a much cooler and easier to use interface. So you can configure the bot quickly and easily, but above all the graphics will be much more intuitive. As soon as I finish developing, I'll upload the code to this repository, so you can simply update.

[spacechicken]

Hey Imusarella,
I found that the only way the current Martingale can be profitable is running it on Quote Strategy - False.
(Currently 11.6% up)

I did like this Martingale Strategy I have attached from a guy that commented previously - it is very sound. @tankiatwee posted before!
Running the current Martingale strategy in any other option makes a large loss. If you be on the low ones - not enough recoup after the losses.

Only.bet.70.percent.xlsx
Martingale.docx

[lmusarella]

You are right, unfortunately the martingale has this limit, if the payout is not high you risk not winning enough, in fact selecting the highest odds could be a valid strategy. Thanks so much for the suggestion!

[spacechicken]

If you code it with that guys strategy for martingale I think that would be a cracker!!!

[tankiatwee]

I tried "Quote Strategy - False" past 2 days and 1 hour ago it went to martingale step 8 because of a known trending period.

Otherwise whether quote strategy is true or false doesn't make much difference because payout changes ever so often on hindsight.

But it would greatly benefit users if Imusarella can implement the extended martingale with throwaway bets of $1 (or the lowest permitted amount set by user as long as gas fee allows; I think that should be 30 cents) to make the bot safe.

[DoctorEam]

Since starting to use these bots I have had the dream of building in machine learning to improve the prediction of bets.
I'm a noob, so all I can use is AI to try to code something like that. The signal strategy could be made way better with different types of signals, e.g. range sentiment profile (or whatever anybody knows to be reliable), but IDK if that would be possible.
A self restarting code in case of errors would be great!

[DoctorEam]

I've heard that for choppy markets the best is to use the quote strategy (lower quote) and for trendy markets going with the higher quote is more effective.

[DoctorEam]

I finally got the bot running and am happy with the way it works...but not the success rate. I set Threshold at 80 and still success rate goes below 50%.

Since martingale is not involved, I like to find out if fellow users have success using the bot in different ways.

For example, I could set reverse betting, or increase bet size together with decreasing daily profit goal so I could be done for the day quickly before overall result gets worse.

Has anyone found out if a particular time period of the day (from what time to what time) yields high success rate?

Any other recommendations are welcome. Please share, thanks.

IME the higher the prediction, the more likely that it fails because it's when the minitrend reverses. A BB/MA strat or so would be more effective perhaps. IDK what would be reliable as 5min timeframes have so much volatility...

[Dennise93293222]

Thanks

[jtruji68]

Hi, i just found this repo. Great work guys, I will give it a try. After all this time, what are the results you have gotten?

[lmusarella]

Hi, if I have to be honest, to get satisfactory results it is useful to use the martingale. Unfortunately it is difficult to exceed 60% success however, with the martingale it is possible (if we have a high enough initial capital) to make profits, combined with the quote strategy it is not bad I'm currently working on a strategy based on Binance and oracles delay, it's promising.

[jtruji68]

thanks for your input. do you think enough money to gef 7 loses max is enough? if the probabilities are >60% with a bit of tweaking that sounds pretty good using martingale or the reverse one

[lmusarella]

Its a good balance, i suggest you tu use the bot in simulation mode, before using real Money. Is rare, but stastistically you can lose 8 times ore more. Il ven 13 ott 2023, 07:51 David Trujillo ***@***.***> ha scritto:

…

thanks for your input. do you think enough money to gef 7 loses max is enough? if the probabilities are >60% with a bit of tweaking that sounds pretty good using martingale or the reverse one — Reply to this email directly, view it on GitHub <#13 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIRDYVFR46QOYHQN2U3ECTLX7DJHLAVCNFSM6AAAAAAXEBNMASVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TENZQGE2DC> . You are receiving this because you were mentioned.Message ID: <lmusarella/Pancakeswap-Prediction-Bot-2023/repo-discussions/13/comments/7270141 @github.com>

[jtruji68]

Thanks @lmusarella , for the new users here, is best to have enough money to tank 9-10 losses min. Is best to get fewer profits than to lose all your bank. That only needs to happen once, remember that.

[jtruji68]

Im back with some testing, I live in UTC-5 and from 8:00am to 5:00pm it works pretty good with this setup:

• SIGNAL STRATEGY
• Threshold 60
• enough budget to last 7 losses
• RPC provider by getblock . com. (had some gas problems using quicknode)
  But at night it gets wayy too many consecutive losses, I think its because lower volume and crabbier pattern at least on BNB.
  Good job dev, Ill keep testing an I'll make a donation.

[mitchbarks]

With your current node at getblock.com, how many seconds out have you set your bot to bet at? The default 35 seconds out from the next round or have you done quicker?

[jtruji68]

Hi, @mitchbarks , sorry for the late reply, I dont browse github too much. As for your question I have to update some things:

• I dont use the getblock.com provider anymore. found"https://bscrpc.com"and "https://bsc-mainnet-rpc.allthatnode.com" to be way better. I dont touch the default timer setting.
• SIGNAL_STRATEGY is not the best one, even if the winrate is positive, you can still lose money, and its no different than gambling.
• The budget you need to survive some really bad streaks should tank at least 9-10 losses, so for example if your budget is $500, your min bet should be no more than $0.47 (you can lose 10 times max), anything else is lost money eventually.
• Imo the best strat is as some user said in the strats thread: QUOTE_STRATEGY, set it the SELECT_LOWER_QUOTE true by default. if you see your winrate is getting negative, switch it to false, but is better to keep it that way because most of the time the lower quotes have better chances, if you are familiar with sports betting you'll understand. The max losses I had in this one was 9, but it was the most profitable one to date. In resume: higher quote = more risk, more profit; lowe quote = less risk, less profit.

[tocacyy]

After a long period of testing in Simulation mode, where this mode does not have 100% accuracy, I say this because I have watched the rounds in this mode, there were situations where the round was win and the bot turned it into a loss. I chose the following strategies:
BET_AMOUNT: 0.25
MARTINGALE Active
ANTI_MARTINGALE: false
INCREMENT_BET_AMOUNT: 2
NUM_ROUNDS_AFTER_DELETE_PENDING_ROUNDS: 2
SELECTED_STRATEGY: 'SIGNAL_STRATEGY'
REVERSE_BETTING: true

With this configuration and with a balance of $150, there were days where it generated up to 10% profit excluding the fee, and it heavily depends on the market conditions