Training your AI

To make your bot work like that follow next steps. Keep in mind that DCA – Merging trading bots don’t produce steady and instant profits, no bots do, they are fighting their way through the market volatility and earning like the picture above. Another good thing about DCA trading bots is that you never sell in loss, you are fixing failed trades by buying lower and averaging down your price of an asset to sell it in profit again once profit target is reached way lower on the charts than before. All you need is time to fix them and use these settings to buy coins at the right time and areas.

*If you are making AI for the first time on Cryptohopper read basic guide here: https://docs.cryptohopper.com/docs/en/Explore%20Features/algorithm-intelligence-ai/

• • Create your new AI, name it, add strategies + trends and change your AI config like it’s shown on the pictures at the beginning of this page or in this video.
  • Train coins inside AI, you can choose more if you wish. (50) BTC, ETH, BNB, LTC, QTUM, ADA, XRP, EOS, XLM, ONT, TRX, ETC, VET, LINK, ZIL, ZRX, BAT, XMR, ZEC, DASH, MATIC, ATOM, ONE, FTM, ALGO, DOGE, CHZ, XTZ, KAVA, BCH, SOL, KNC, COMP, ZEN, SXP, MKR, STORJ, CRV, DOT, SUSHI, EGLD, RUNE, UNI, AVAX, AAVE, NEAR, GRT, XNO, SANTOS, FXS
    

• • When it’s done training you should also train them on real trades using your hopper with paper money to make it smarter and fix unrealistic results from the backtesting engine. Download Star AI Paper Trainer template and select your exchange in baseconfig settings. (You will notice the Star AI trainer template has more coins than we have in real trading Star AI template, you will have them pre-trained and ready to use if you ever decide to switch them and trade other coins)

Make sure you select your named AI under strategies in baseconfig settings when loading template. (not shown in video above)

• Your paper bot should be running Star AI Paper Trainer template for 1-2 weeks. Idea is to backtest and train your AI on live trades which are much more accurate than history backtesting system. Backtesters are never 100% reliable and feed too much inaccurate information to AI.  Template is designed to score and validate buys exactly the way AI system does it. If backtesting is not accurate, train it on real trades right?
• Stats and results on paper bot won’t make sense, don’t worry about that, it’s really designed to just score and train your AI for now.

(You can build your AI and train it on paper with any subscription active. You don’t need to buy HERO yet (yes, even Explorer). Buy or upgrade your subscription when you are done with pre-training and put AI on live bot.)

Most of CH users say AI doesn’t work? You will be surprised how smart it can become.

*Pre-training refers to training a model with one task to help it form parameters that can be used in other tasks.

*Star AI Paper Trainer. We are training AI on all strategy signals. AI remembers every signal and sort them into trend categories with success.

When you are done paper training after 1 or 2 weeks, you will see that strategies got scored down by 5-10-15%. Don’t worry this means it became smarter and fixed fake results. It’s finally time to start trading Star AI template for real. I advise you to start with ~90%+ minimum Buy Score of AI at the very beginning or even higher to make sure it’s really settled for a day or two, then you can lower it.

We are taking advantage of the scoring system and trade with the market trend this way for each individual coin. Check where is the score limit to avoid downtrend and other bad buys in your AI by clicking on strategy inside AI -> best markets (GIF image below). Look for the highest winning scores of coins on neutral/downtrend/uptrend markets and remember them for that trend. Trade above downtrend/neutral scores and get just uptrend buys for example. I suggest you trade above neutral/downtrend scores in long bear market to really jump in and out fast and don’t get drawn down. In a long bull/sideway market you can easily have a lower buy score. 90-80% and include neutral – sideway market buys.

*Example – if your highest score on coins for ‘trend neutral %’ is 88% and your highest score for ‘trend down %’ on coins is 84%, trade just >90% buying score to get only uptrend buys or maybe set it at 86% to include sideway-neutral market signals as well.

Once you get just 1-10 trades a day you know you are hard limiting AI and don’t let it buy downtrend coins which is your goal. If you want more trades slowly slide down the buy score button to get more buys and watch the performance. You can quickly regret it so use it with caution. Keeping it on high buys score you make even DCA – Merging buys getting triggered at amazing spots and skip all bad buying signals.
After 2 or more months of trading, downtrend score limits usually fall below 80% so trading above 81% is always highly advisable. Yes, to cut all things short. Set it on ~85% and you don’t need anything else except v1 triggers turned on (default) when your AI settles down with live scores since AI is already trading with the trend if you just use a good scoring system and proper training like this. Have fun!

Tips: (join Discord community or contact us to discuss them)

• Turn on v2 ema if your AI is relatively new and you still want to have that extra safety guard on your bot and trade inside the best optimized EMA trend for this markets. Pointing out to AI’s trained less than 2 weeks or trained just in a strong uptrend market where all of the trades profit. You want it to experiance some downtrend market and crashes for the best training of your AI.
  

• When AI is settled down after 1,2 weeks – should be enough –  and you see strategies got scored down by 10-15%, it actually became smarter and knows how to trade now.  Start your real bot with just v1 triggers turned on (default) on Star Ai template and have a high buy score of ~85-90%+ in the template which will trade by the trend for each individual coin on this high score so there’s no need for v2 EMA triggers anymore when Ai is trained.
  

• In long downtrend markets I suggest you trade on ~85% + buy score, to catch just uptrend buys and avoid neutral/downtrend buys. Neutral or sideway trend buys usually ‘break down’  after consolidating and not ‘break out’ to the upside after some time most likely.
  

• In long uptrend charts you can lower buy score to 80% easily, which means you will take some trades  in neutral and downtrend as well. You can see more info what are your scores on coins for different market trends inside AI – click strategy, I prefer Scalp Star [Bull] – best markets and see percentages. We talked about this before, check that GIF picture up there.

• Set and forget AI should be set on high ~85% and just let it work without changing.
  

• V1 triggers are based on percent change candle, when I talk about V2 these are triggers based on trends (EMA).

• If you have a spare paper bot idling it’s advised to keep running Star AI Paper Trainer parallel with your real bot later. The idea is to have a real bot on high buy score to get and trade just the best signals from AI and keep paper bot active on just 51% buy score to constantly score all signals and keep validating them. This can make bad coins good again in the future if market turns up to be good for them and slowly score trades back. This way you basically create live backtester on paper bot and trade just good signals on real bot.
  (Don’t run paper trainer for 1 year nonstop heh.. Don’t over-spam the scoring system. Personally, I like to turn on paper trainer here and there to feed my AI more info every 3 months .. for 1 week and then turn it off again. Live trading AI always turned on of course.)
  

• DCA/Merge trading is usually a long-term investment. You won’t see 1% daily gains but you will see profits in the longer term. Invest money and keep trading it for 3..6 months and get out your profits. Star AI trades and profits on small market bounces every day, you don’t need huge uptrend and all-time high prices to take out profits.That’s simple holding not trading, these strategies are trying to fight the trend by taking small profits on market bounces, waves and good volatility.

• These strategies never buy the top or signal on the way up. It’s designed like that for many reasons to make your bot run fluently through any market. We buy on pullbacks, corrections, consolidations, support lines, bull flags and trend reversals.

• We trade using merging system, which is your wallet and user friendly. Double down, triple down way of trading averages down your trades faster, but not many people can afford this way of trading. We are trading small amounts and keep investing into markets by merging on more markets at once, this goes a long way and you are not being stressed about red positions since you invest pocket money little by little in every coin.

• Many users think AI doesn’t work. We studied AI for a long time and know all the little details and tricks of this system. 90% of strategies on CH will never work well in AI, there are reasons why and why strategies never self-train inside AI, we tested it for more than a year to fully understand it and this knowledge is our secret. We made sure to make the best performing AI.
  

Choosing the correct markets

Choosing the correct coins and tokens is the most important part of trading. We target markets that are range bounding for a long time already with good volatility, volume, liquidity and have a strong name in the crypto world. You want to be trading cryptocurrencies that people use for daily transactions and sending money across the globe, being decentralized, which was the original idea why cryptocurrencies exist.

A range-bound market is one in which price bounces between a specific high and a low price. The market movement could be classified as horizontal, ranging, or sideways. This means you don’t need to have strong Bull or Bear markets to profit. You will continuously trade to profit on spikes and bounces in the market. All you need is time and patience, bot will fix bad trades by Merging and DCAing on lower prices producing passive money.

Another interesting thing that not many investors even understand. What you see on the picture below is called math behind saying “losses work geometrically against you” which Mark Minervini writes in his books.
You can win 100,500% trades, but you can not lose more than 100%. Your value of an asset is 0$ at -100%, you can not be -200% in loss. What’s the point? The thing is no matter how scary market crash is big market cap coins like BTC usually reverse bear market charts after -50-60% and other higher market cap alts probably -60-100%. If you do just 1 or 2 good DCA’s on -50% and -80%, I can guarantee you will get out of your position and chase for the profit again. That’s why people love DCA trading. Emotions are the problem and get on your way usually, let bots do the work and of course settings must be set correctly to trade through the markets fluently.

Coins

These are the top 15 carefully selected coins right now for Star AI. Not many coins or tokens exist that work this way, so finding them and trusting the project plus looking at other factors like 500M $ + monthly / 3 month volume and so on is tricky. If you do find more we always discuss them on the Discord community (forum) and update them if needed. Please note that if any of this coins doesn’t have a good volume on your exchange, don’t trade them, replace some of the pairs with higher market cap coins (top 50). For example SANTOS, CELO, SXP might not have a good volume on Kucoin, but all of these coins do on the Binance.
XMR, DASH, XTZ, EOS, ETC, XRP, TRX, AAVE, UNI, ZIL, XLM, ZEC, BAT, KAVA, ZRX