Strategies: TA (ema+macd) and Trust/Distrust #285

DeviaVir · 2017-06-14T08:35:01Z

This TA is a copy of the work @Fjuxx has been doing, since it's a separate strategy, I don't think there's a harm in including it already.

The trust/distrust can be described as follows:

the idea is that you trust the coin to keep growing, but that you distrust it when it comes to peaks. simply put: you tell it what the min sell percentage is (what percentage the current price should be above the starting, or buy-in, price), you tell it at what percentage below the highest point you wish to sell it at, and you tell it at what percentage above the deepest point you wish to buy at. All reset after an action.

Trust/Distrust is best used with (expected) growing altcoins. Trust/Distrust is nice if you don't want the bot to ever make a negative trade (although you can configure the sell_threshold_max which is the percentage below at which the bot will panic sell)

Here are some results from trust/distrust:

$ ./zenbot.sh sim poloniex.DGB-BTC --strategy=trust_distrust --sell_threshold=3 --sell_min=2 --buy_threshold=2 --period=28m --sell_threshold_max=0 --poll_trades=5000 --order_adjust_time=5000 --greed=350 --currency_capital=1400 --asset_capital=0 --days=30

end balance: 7263.64799015 (418.83%)
buy hold: 3870.28985507 (176.45%)
vs. buy hold: 87.68%

vs a not super fast growing market (these params should be further tuned):

$ ./zenbot.sh sim gdax.ETH-EUR --strategy=trust_distrust --sell_threshold=3 --sell_min=2 --buy_threshold=1 --period=28m --sell_threshold_max=10 --poll_trades=5000 --order_adjust_time=5000 --currency_capital=1400 --asset_capital=0 --days=14

end balance: 2005.97032216 (43.28%)
buy hold: 2120.40945035 (51.46%)
vs. buy hold: -5.40%

another not fast growing market:

$ ./zenbot.sh sim gdax.BTC-EUR --strategy=trust_distrust --sell_threshold=3 --sell_min=2 --buy_threshold=1 --period=28m --sell_threshold_max=10 --poll_trades=5000 --order_adjust_time=5000 --currency_capital=1400 --asset_capital=0 --days=14

end balance: 1563.41699875 (11.67%)
buy hold: 1640.53128599 (17.18%)
vs. buy hold: -4.70%

(trend_ema balance for the above market: end balance: 1358.10164583 (-2.99%))

Fjuxx · 2017-06-14T09:18:13Z

Please note that the TA is still WIP, i'm planning to work on it a bit further at the end of this week.

as you can see, with the MACD it processes ALL data to TAlib. with the EMA only the last x periods are fed.

As well i found that at leased for MACD the data seems to be 1 period off and i dont know why yet. this needs more testing.
(it could be that the difference is coming from the fact that the bot is calculating it realtime and the internet graphs are calculating it backwards.

DeviaVir · 2017-06-14T09:24:33Z

@Fjuxx noted! This is why I added them as separate ta_ strategies, so we are not overwriting anything while this is WIP.

Fjuxx · 2017-06-14T09:28:43Z

Perfect, when i start again i will work from this branch (or if it is merged from master ofc)

simonwestyn · 2017-06-15T21:21:47Z

package.json

@@ -32,6 +32,7 @@
    "semver": "^5.3.0",
    "sosa_mongo": "^1.0.3",
    "timebucket": "^0.4.0",
+    "talib": "^1.0.3",


FYI, Talib doesnt work on Windows:
npm ERR! notsup Unsupported platform for talib@1.0.3: wanted {"os":"darwin,linux","arch":"any"} (current: {"os":"win32","arch":"x64"})

Although, yes, that sucks, TA-Lib has many indicators and many financial functions that might be of use. This could be a case where I'd say that we drop Windows support.

Just wanted to let you know. But I agree, VM on windows is easy

arpheno · 2017-06-16T13:39:04Z

I'm noticing major slowdown in performance for the two ta_ strategies. Are you observing the same behaviour?

tycho · 2017-06-18T02:24:41Z

extensions/strategies/trust_distrust/strategy.js

+    description: 'Sell when price higher than $sell_min% and highest point - $sell_threshold% is reached. Buy when lowest price point + $buy_threshold% reached.',
+
+    getOptions: function () {
+      this.option('period', 'period length', String, '30m')


You're missing a min_periods value here, which causes a NaN when trying to calculate the number of days to backfill here.

I feel like this min_periods shouldn't be part of the strategy, as it's needed in trade.js... but I added it, as I see that all other strategies, except for SAR, have also added it (but don't actually do anything with it).

Good catch!

nao0811ta · 2017-06-18T05:21:31Z

I have same problem.so reset to previous commit.
Signal is to null.

DeviaVir · 2017-06-18T17:01:27Z

@arpheno yes, those are definitely WIP at this time. This is also the reason that they've gotten their own strategy name instead of replacing the current ema and macd.

DeviaVir · 2017-06-18T17:02:40Z

@nao0811ta I'm not sure what you're referring to, what signal is to null?

node-gyp and node v8.x are not friends, stick to boron for now Rename talib's MACD (#2) * Rename talib's MACD so it doesn't overlap with original macd

DeviaVir · 2017-06-20T00:02:05Z

extensions/strategies/trust_distrust/strategy.js

+      if (s.trust_distrust_lowest > s.period.high) {
+        s.trust_distrust_lowest = s.period.high
+      }
+    },


note to self: might want to consider calculating EMA/MACD (both?) to check if we are in an up or down trend, and advise buy/sell accordingly.

currently we sell when we drop a certain percentage under the top, but if EMA/MACD says we are in an upward trend, it doesn't make sense to sell.
currently we buy when we grow a certain percentage above the deepest point, but if EMA/MACD says we are in a downward trend, it doesn't make sense to buy.

Hi @DeviaVir, did you manage to implement this idea of yours?

nibzo1 · 2017-07-03T21:19:41Z

@DeviaVir whats the status of this PR, seems like a good strategy for taking Max profit.

DeviaVir · 2017-07-03T21:24:07Z

@nibzo1 could definitely still use some more eyes, tweaks and other things. It also breaks the Windows version (by trying to install ta-lib) so I'm not merging unless I have contributor agreement at those pros and cons.

nibzo1 · 2017-07-03T21:31:53Z

Ok sure thing I might bull changes and give it a test run, I'm not a windows user so I wont be able to assist there.

I can definitely do some sim's and even some live testing against the Kraken exchange and report back any odd behavior.

bchadek · 2017-07-13T04:20:44Z

I think the two ta_ strategies are so slow because the arrays used in the corresponding lib files grow rather large. Perhaps making the s.marketData just what is needed (close), instead of a multidimensional array, then slicing the last n that are needed into a new working array var will help

txssseal · 2017-07-20T21:44:29Z

extensions/strategies/trust_distrust/strategy.js

@@ -0,0 +1,136 @@
+var z = require('zero-fill')


Could you comment what exactly this strategy does at the top? What you wrote in the description of this pr would be really awesome.

DeviaVir · 2017-07-28T10:20:07Z

My biggest mental blocker for this was Windows support. Which it looks like we have already broken, so I'm inclined to just merge this now and allow others to play with this in master.

…os8f-master * 'master' of https://github.com/carlos8f/zenbot: Improve command help for train command (DeviaVir#436) Support for missing win/loss and error rate from output (DeviaVir#426) Fix docker builds for forex.analytics (DeviaVir#424) Minor fixes for Quadriga trading API (DeviaVir#421) xmpp for trading alarms (DeviaVir#333) Strategies: TA (ema+macd) and Trust/Distrust (DeviaVir#285) Fixed docker build (DeviaVir#349) updated c.default_selector to c.selector in README.md (DeviaVir#418) added check for message before doing anything with it (DeviaVir#412) FIX: Properly check for unknown indicators in forex_analytics (DeviaVir#408) Added support for the BTCe exchange (DeviaVir#388) Add strategy: forex.analytics, an genetic optimization algorithm for TA-lib stats (DeviaVir#389) # Conflicts: # .gitignore # commands/sim.js # extensions/exchanges/bittrex/exchange.js # extensions/exchanges/quadriga/exchange.js # lib/engine.js # scripts/auto_backtester/backtester.js

node-gyp and node v8.x are not friends, stick to boron for now Rename talib's MACD (DeviaVir#2) * Rename talib's MACD so it doesn't overlap with original macd

DeviaVir requested review from carlos8f and egorbenko June 14, 2017 08:35

Repository owner deleted a comment from RyanF109 Jun 14, 2017

simonwestyn reviewed Jun 15, 2017

View reviewed changes

tycho reviewed Jun 18, 2017

View reviewed changes

Strategies: TA (ema+macd) and Trust/Distrust

9790c32

node-gyp and node v8.x are not friends, stick to boron for now Rename talib's MACD (#2) * Rename talib's MACD so it doesn't overlap with original macd

DeviaVir commented Jun 20, 2017

View reviewed changes

Merge branch 'master' into deviavir/strategies

9ea0ccb

txssseal reviewed Jul 20, 2017

View reviewed changes

DeviaVir merged commit 255a057 into DeviaVir:master Jul 28, 2017

DeviaVir deleted the deviavir/strategies branch July 28, 2017 10:20

DeviaVir mentioned this pull request Jul 28, 2017

Update documentation for Windows users #420

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Strategies: TA (ema+macd) and Trust/Distrust #285

Strategies: TA (ema+macd) and Trust/Distrust #285

DeviaVir commented Jun 14, 2017 •

edited

Loading

Fjuxx commented Jun 14, 2017

DeviaVir commented Jun 14, 2017

Fjuxx commented Jun 14, 2017

simonwestyn Jun 15, 2017 •

edited

Loading

DeviaVir Jun 16, 2017

simonwestyn Jun 16, 2017

arpheno commented Jun 16, 2017

tycho Jun 18, 2017

DeviaVir Jun 18, 2017

nao0811ta commented Jun 18, 2017

DeviaVir commented Jun 18, 2017 •

edited

Loading

DeviaVir commented Jun 18, 2017

DeviaVir Jun 20, 2017 •

edited

Loading

LuisAlejandro Dec 15, 2020

nibzo1 commented Jul 3, 2017

DeviaVir commented Jul 3, 2017

nibzo1 commented Jul 3, 2017

bchadek commented Jul 13, 2017 •

edited

Loading

txssseal Jul 20, 2017

DeviaVir commented Jul 28, 2017

Strategies: TA (ema+macd) and Trust/Distrust #285

Strategies: TA (ema+macd) and Trust/Distrust #285

Conversation

DeviaVir commented Jun 14, 2017 • edited Loading

Fjuxx commented Jun 14, 2017

DeviaVir commented Jun 14, 2017

Fjuxx commented Jun 14, 2017

simonwestyn Jun 15, 2017 • edited Loading

Choose a reason for hiding this comment

DeviaVir Jun 16, 2017

Choose a reason for hiding this comment

simonwestyn Jun 16, 2017

Choose a reason for hiding this comment

arpheno commented Jun 16, 2017

tycho Jun 18, 2017

Choose a reason for hiding this comment

DeviaVir Jun 18, 2017

Choose a reason for hiding this comment

nao0811ta commented Jun 18, 2017

DeviaVir commented Jun 18, 2017 • edited Loading

DeviaVir commented Jun 18, 2017

DeviaVir Jun 20, 2017 • edited Loading

Choose a reason for hiding this comment

LuisAlejandro Dec 15, 2020

Choose a reason for hiding this comment

nibzo1 commented Jul 3, 2017

DeviaVir commented Jul 3, 2017

nibzo1 commented Jul 3, 2017

bchadek commented Jul 13, 2017 • edited Loading

txssseal Jul 20, 2017

Choose a reason for hiding this comment

DeviaVir commented Jul 28, 2017

DeviaVir commented Jun 14, 2017 •

edited

Loading

simonwestyn Jun 15, 2017 •

edited

Loading

DeviaVir commented Jun 18, 2017 •

edited

Loading

DeviaVir Jun 20, 2017 •

edited

Loading

bchadek commented Jul 13, 2017 •

edited

Loading