Portfolio of strategies
A. Christian Silva
3/1/2024
Summary
This document introduces three systematic trading strategies for ETH and/or BTC that require little human intervention: TSM, LTSM, and MLM (L or S). TSM is the only strategy with a brief track record of several months. Therefore, for consistency, I show simulations for all models. LTSM is essentially an extension of TSM and therefore quite stable and ready to go live. MLM is the most sophisticated strategy, and research on it is still ongoing.
Time-series momentum (TSM)
Strategy
TSM is a systematic trading strategy that capitalizes on strong directional crypto asset movements of the order of few months. The strategy is designed to perform equally well in bull or bear crypto markets by going long or short ETH and/or BTC.
The core algorithm is based on extensive academic research into price momentum strategies. The algorithm measures momentum utilizing moving averages of trailing prices build in order to provide a convex payoff and reduce turnover. The risk control algorithm utilizes risk-parity to archive an approximately constant volatility target of 10% annually.
Trades ETH and BTC.
Long/Short (trading perpetuals).
No leverage with 10% target volatility which can be adjusted.
Performance
Live
TSM has been trading under the name of “Beta+” at Teahouse finance. Since 09/2023 it has an AUM of approximately $240.000. The performance is in line with simulation for the same period. Please see link for performance and other details. However, please note that Beta+ implementation details (using AMM etc) has a minimum impact on performance.
Simulated
The simulated strategy from 01/2016 to 01/2024 shows a yearly return of 13.5%, a standard deviation of 10.5% and a daily 1% VaR of -2.3%.
| Statistics (2016-2024) | |
| TSM | |
|---|---|
| Annualized return(%) | 13.50 |
| Annualized volatility(%) | 10.50 |
| Annual Sharpe (Rf=0) | 1.29 |
| Max drawdown(%) | 18.00 |
| Day corr w/BTC(%) | 2.00 |
| TSM | |||||||||||||
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2020 | 0.0% | 2.4% | 1.6% | 1.4% | −0.9% | −2.5% | 9.0% | 1.8% | −0.5% | 0.7% | 7.8% | 3.9% | 27.0% |
| 2021 | 4.7% | 0.2% | −3.2% | 2.0% | −0.9% | 0.1% | −0.1% | 1.6% | −0.1% | −0.1% | −1.3% | 1.4% | 4.3% |
| 2022 | 1.5% | −2.7% | 0.2% | 0.7% | 5.2% | 9.1% | −1.9% | 1.3% | −0.8% | −0.3% | −4.2% | −0.1% | 7.8% |
| 2023 | 8.4% | −1.7% | −7.8% | −0.7% | −2.6% | −0.1% | −1.0% | 0.6% | −1.0% | 0.0% | 2.9% | 1.3% | −2.5% |
| Simulated returns that include transaction costs but do not include management fees and/or performance fees. The simulation also does not include interast rate on cash reserves. | |||||||||||||
TSM compared to BTC. BTC is adjusted to have the same TSM volatility.
TSM compared to BTC. BTC is adjusted to have the same TSM volatility.
Long time-series momentum (LTSM)
Strategy
Long only strategy that leverages TSM algorithm to mitigate bear markets.
Trades ETH and BTC.
Long only. Algorithm dynamically controls the long exposure to be between 20% and 100% long.
No leverage: we can trade spot on 1inch for example.
Adapts the current Beta+ algorithm to size the long position and to control risk.
Performance
Live
Live performance can be found at Teahouse under the name Beta+Long. Please see documentation for implementation details.
Simulation
Transaction cost of 5 bp per trade (round trip 10 bp). This value comes from our experience with TSM.
Volatility is an average of 55% per year. It is less than long only due to timing mechanism.
AUM = $200,000 : $100,000 is invested into ETC and $100,000 is invested in BTC.
Interest rates are taken to be zero: cash positions do not accrue interest.
| Statistics (2016-2024) | ||
| LTSM | BTC | |
|---|---|---|
| Annualized return(%) | 94.20 | 77.5 |
| Annualized volatility(%) | 54.30 | 70.4 |
| Annual Sharpe (Rf=0) | 1.73 | 1.1 |
| Max drawdown(%) | 53.00 | 83.0 |
| Day VaR 99% (%) | -10.30 | -15.2 |
| Day corr w/BTC(%) | 74.00 | NA |
| LTSM | |||||||||||||
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2020 | 24.9% | 9.2% | −5.7% | 22.8% | 0.7% | −6.8% | 34.5% | 8.4% | −13.7% | 11.5% | 50.1% | 21.0% | 278.1% |
| 2021 | 41.7% | 21.8% | 15.3% | 9.5% | −1.6% | −1.8% | 2.3% | 23.2% | −6.7% | 35.6% | −4.7% | −4.2% | 206.3% |
| 2022 | −4.6% | 0.0% | 6.6% | −12.1% | −4.7% | −9.5% | 22.6% | −4.6% | −7.8% | 7.7% | −9.8% | −3.8% | −22.3% |
| 2023 | 28.9% | −3.5% | −2.7% | 1.9% | −5.9% | 4.8% | −3.7% | −6.6% | 0.6% | 11.3% | 10.5% | 7.2% | 45.0% |
| Simulated returns that include transaction costs but do not include management fees and/or performance fees. The simulation also does not include interast rate on cash reserves. | |||||||||||||
Machine learning model: long (MLML)
Strategy
Strategy uses pattern recognition and machine learning to identify up trends.
Strategy is long or flat and it can have holding period of few hours.
IMPORTANT! Strategy is under active development: performance as well as other stats will change.
Performance (simulation)
- Trading ETH only.
- Long or flat but never short.
- TC = 5bp per trade.
- Interest rates are taken to be zero: cash positions do not accrue interest.
- AUM = $200,000.
| Statistics (2016-2024) | ||
| MLML | BTC | |
|---|---|---|
| Annualized return(%) | 134.8 | 77.5 |
| Annualized volatility(%) | 64.3 | 70.4 |
| Annual Sharpe (Rf=0) | 2.1 | 1.1 |
| Max drawdown(%) | 56.0 | 83.0 |
| Day VaR 99% (%) | -7.8 | -15.2 |
| Day corr w/BTC(%) | 51.0 | NA |
| MLML | |||||||||||||
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2020 | 32.7% | 26.7% | −6.7% | 31.8% | 0.3% | −4.8% | 39.0% | 31.3% | 3.0% | 1.1% | 40.5% | 9.3% | 483.5% |
| 2021 | 58.1% | 19.2% | 16.3% | 24.2% | 12.7% | 4.3% | 6.3% | 14.1% | −2.0% | 19.3% | 15.7% | −12.8% | 357.2% |
| 2022 | −19.4% | −3.2% | 9.7% | 2.3% | −2.6% | −12.2% | 31.0% | 9.5% | −5.4% | 9.3% | 1.9% | −6.6% | 5.8% |
| 2023 | 18.4% | 7.2% | 0.3% | 1.9% | 0.3% | 9.0% | −2.1% | −6.7% | −2.4% | 2.8% | 10.6% | 3.3% | 48.5% |
| Simulated returns that include transaction costs but do not include management fees and/or performance fees. The simulation also does not include interast rate on cash reserves. | |||||||||||||
Machine learning model: long/short (MLMS)
Strategy
Strategy uses pattern recognition and machine learning to identify trend and mean-reversion periods.
Strategy is long/short and it can have holding period of few hours.
IMPORTANT! Strategy is under active development: performance as well as other stats will change.
Performance (simulation)
- Trading ETH only.
- Long/Short.
- TC = 5bp per trade.
- Interest rates are taken to be zero: cash positions do not accrue interest.
- AUM = $200,000.
| Statistics (2016-2024) | ||
| MLMS | BTC | |
|---|---|---|
| Annualized return(%) | 163.90 | 77.5 |
| Annualized volatility(%) | 68.10 | 70.4 |
| Annual Sharpe (Rf=0) | 2.41 | 1.1 |
| Max drawdown(%) | 52.00 | 83.0 |
| Day VaR 99% (%) | -10.00 | -15.2 |
| Day corr w/BTC(%) | 6.00 | NA |
| MLMS | |||||||||||||
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2020 | 26.9% | 31.5% | 26.6% | 0.9% | 1.1% | 0.0% | 32.2% | 47.1% | 30.6% | −10.6% | 34.8% | 9.6% | 632.7% |
| 2021 | 22.5% | 36.0% | 6.0% | 2.7% | 14.6% | 30.8% | −1.4% | −2.9% | 1.1% | −2.1% | 30.7% | −12.6% | 194.5% |
| 2022 | −19.1% | −21.8% | 1.6% | 19.2% | 10.8% | 30.6% | 23.7% | 26.6% | 3.1% | −0.1% | 26.0% | −6.2% | 111.0% |
| 2023 | 10.5% | 16.9% | −12.5% | 4.7% | 3.3% | 15.1% | 2.9% | −6.2% | −6.7% | −0.7% | 7.3% | 2.0% | 37.8% |
| Simulated returns that include transaction costs but do not include management fees and/or performance fees. The simulation also does not include interast rate on cash reserves. | |||||||||||||
Disclaimers
It is important to note that simulations are not real returns and therefore can only serve as an approximate view of the risk/return profile of a given strategy. Finally, it is fundamental to stress that past performance is not indicative of future results, and any historical returns presented should not be taken as a guarantee of future performance. Investors should carefully consider their financial objectives, risk tolerance, and investment horizon before investing.