Ethereum Star MT5
- Experts
- Kristian Kafarov
- Versione: 1.1
- Aggiornato: 27 aprile 2024
- Attivazioni: 7
Trend following expert on Ethereum with live performance results (see screenshots) and MT5/Binance trading mode.
Introduction
Have you ever wonder why is it so hard to accumulate impressive capital using expert advisors? The key feature here is reinvestment. You need to constantly increase your deposit so you could receive greater income. And compound that income to your deposit again. Then the power of compound interest will increase your capital exponentially over time.
To do that you can choose between two types of expert advisors:
- Martingale/grid experts with unlimited losses or experts with losses much greater than profits.
- Experts with fixed losses which is less or comparable to profits.
- In the first case you have to withdraw all your profit because sooner or later the expert will burn your entire deposit. Assuming typical performance of grid experts, it will take a year or more to simply return 100% of your investment.
- In the second case you can compound your profit but it's difficult to achieve the same stability of performance compared to grid systems because such experts always tend to get overfitted to the past prices.
I've been stuck with this dilemma for a long time until I've created this expert.
Algorithm
- On the one hand, it uses some adaptive conditions to enter the trend with a relatively short SL, follow that trend with trailing SL and exit or reverse at the end of the trend. These conditions adapt to the market situation.
- On the other, it has some filters to avoid multiple losses during the flat phase, which were tuned based on my original technique that I call "an optimization without optimization". It was inspired by several research articles ([1], [2], [3]) where I found the answer why most of the optimization techniques used in algotrading are fail. This approach allows to reduce an overfitting to minimum.
Main properties of the expert:
- No grid, no martingale, not holding loss positions for a long time.
- Each deal has a fixed SL that can only be shortened.
- Uses trailing SL to maximize profit from big trends.
- Requires no optimization, use default settings adjusted to your deposit.
- Shows stable historical results during both bull and bear crypto trends, high and low volatility, from 2018 to 2024.
- Shows historical performance (rec.factor=29.5 with a fix.lot) compared to the best grid systems without dangerous trading methods.
- Demonstrates live performance similar to test results (see screenshots).
- Comply with FIFO rule.
- Compatible with both netting and hedging accounts.
- Minimal deposit is 100$.
- The only recommended symbol is Ethereum. Can also be used with BNB but performance is worse.
- The only recommended timeframe is M15.
- Not sensitive to slippages and not critically sensitive to comission.
Parameters:
| Parameter | Default value | Description | Comment |
|---|---|---|---|
| WorkTimeframe | PERIOD_M15 | timeframe for trading via MT5 | Preferrable timeframe is M15. |
| TradeBinance | false | trade on Binance | False for trading via MT5. True for trading directy on Binance Futures. |
| ApiBinance | your Binance account generated API | Required if TradeBinance=true. | |
| SecretBinance | your Binance account generated secret | Required if TradeBinance=true. | |
| SymbolBin | ETHUSDT | Binance instrument | Symbol name from Binance futures. |
| TimeFrameBin | 15m | Binance timeframe | Preferrable is 15m. Others: 1m/3m/5m/15m/30m/1h/2h/4h/6h/8h/12h/1d/1w. |
| DigitsBin | 2 | decimals of Binance instrument | ETHUSDT has 2. |
| Lot | 4 | base lot | In practice, I use no more than 1 lot for every 1000 of balance |
| PriceNorm | 3000 | price for a volume = Lot | Needed to eliminate the impact of price level on risk. 3000 means it will trade 4 lots when ETH price is 3000, 3 lots when price is 4000, 6 lots when price is 2000 etc. It will keep the size of position (price x lot = 12000) same for every price level. |
| AdjustLotByBalance | false | increase lot with balance | |
| BalanceNorm | 8000 | balance for a volume = Lot, when AdjustLotByBalance=true | If Lot=4 will trade 2 lots with 4000 balance, 3 lots with 6000 balance etc. |
| MaxSpread | 0.2 | max spread in % of price | On Binance ETHUSDT futures spread is usually 0.003% of price. |
| AllowBuy | true | buys allowed | |
| AllowSell | true | sells allowed | In case you'll want to use one side trading. I always use both sides even during strong bull market because there's a chance to get additional profit on sudden rapid corrections + you never know when the trend will revert. |
| CloseAtFriday | false | close all positions at friday evening | |
| CloseAtFridayHour | 22 | hour to close all at friday (GMT) | In case your broker doesn't give a quote on weekend. The best results achieved when trading 24/7. |
Testing
All tests were performed with Binance ETHUSDT spot prices. You can download this data from Binance only using API keys and Python script to transform prices to MT5 format. I provide this script here. Otherwise, you can simply use the MT5-compatible history file prepared by this script.
Create a custom symbol and fill it up with the history from that file as shown here.
Tester should have following settings:
- Period: M15
- Dates: 2018.01.01 to 2024.04.01
- Delays: zero latency, ideal execution
- Modelling: open prices only (no need for better accuracy because the expert ensures enter price to be at the bar open price or puts a limit order)
- Deposit: 8000 USD
- Leverage: 1:500
- Comission: 0.05% for both sides (USDT taker) which is the largest comission on Binance futures (in practice, comission will be between 0.05% and 0.02% because in some cases expert uses limit orders). Set it from my comission config file.
Test results by year and risk mode:
| Dates | Risk | Net profit | MaxDD | MaxDD% | Rec.factor (net.prof./maxDD) | Deals | Sharp |
|---|---|---|---|---|---|---|---|
| year 2018 | fix.lot=4 | 28 811 | 3532 | 34% | 8.2 | 177 | 4.00 |
| year 2019 | fix.lot=4 | 5522 | 3931 | 34% | 1.4 | 208 | 0.94 |
| year 2020 | fix.lot=4 | 21 390 | 2108 | 19% | 10.1 | 194 | 3.95 |
| year 2021 | fix.lot=4 | 25 523 | 3505 | 29% | 7.3 | 190 | 4.25 |
| year 2022 | fix.lot=4 | 22 405 | 3514 | 17% | 6.4 | 216 | 4.02 |
| year 2023 | fix.lot=4 | 11 042 | 2353 | 21% | 4.7 | 220 | 2.28 |
| 1.1.2018-1.4.2024 | fix.lot=4 | 116 169 | 3931 | 34% | 29.5 | 1263 | 2.32 |
| 1.1.2018-1.4.2024 | lot=4 for every 20k bal. | 1 796 594 | 168 296 | 18% | 10.7 | 1263 | 3.27 |
| BTC 1.1.2018-1.4.2024 | buy and hold for 8000$ | 32 400 | 29 409 | 77% | 1.1 | 1 | |
| ETH 1.1.2018-1.4.2024 | buy and hold for 8000$ | 30 640 | 40 410 | 89% | 0.76 | 1 |
Trading
I personally use this expert to trade on Binance futures. Since it was developed with spot prices it uses spot prices to analyze market and decide what to do but trades futures because of the lower fees and available leverage. It increases performance compared to analyzing futures prices.
See my simple guide how to run this expert in Binance trading mode.
Comparison between ETHUSDT Binance spot and Alpari Real account ETHEREUM CFD (0.07% comission + spread) performance from 1.10.2021 to 1.4.2024, fixlot=4:
| symbol | comission | net profit | maxDD | rec.factor | deals | sharp |
|---|---|---|---|---|---|---|
| ETHUSDT Binance | 0.05% | 38 100 | 3500 | 10.9 | 567 | 2.17 |
| Alpari Real ETHEREUM CFD | 0.07% + spread | 33 400 | 3350 | 9.2 | 573 | 1.41 |
Of course you will need a VPS server to properly run this expert. I use it with Metatrader VPS.
Results
The results I got on my futures account with this expert absolutly satisfied me. I'm actively continue to trade it and involve more capital. I cannot provide link on my Binance live monitoring because of the Market rules, so I leave only an anonymized screenshot so you could evaluate the performance. For the same reason I provide no MQL signal with my expert: trading conditions that MT5-compatible brokers provide to trade Ethereum CFD are slightly worse than Binance futures.
Feel free to ask me any questions in PM after purchase.
Have a good trade!
References
[1] D. Bailey, J. Borwein, M. López de Prado and J. Zhu, The probability of backtest overfitting, 2013, working paper.
[2] D. Bailey and M. López de Prado, The Sharpe ratio efficient frontier, Journal of Risk 15(2) (2012), 3–44.
[3] Bailey, D., J. Borwein, M. L´opez de Prado and J. Zhu, “Pseudo-mathematics and financial charlatanism: The effects of backtest over fitting on out-of-sample performance,” Notices of the AMS, 61 May (2014), 458–471.
