Please complete this form to download a copy of the whitepaper
Any backtest is only as good as the information put into it. And yet when it comes to trading costs it is often assumed that an approximation is accurate enough to account for real world trading frictions. In reality, the impact of trading costs can be similar in magnitude, or even outweigh, the systematic premia they aim to capture. Additionally, trading costs assume greater significance in a low-return environment as they effectively represent a greater proportion of potential return available.
Our whitepaper explores in detail how to construct investment strategies that accurately incorporate trading costs into the backtesting process.
When the hidden costs of a strategy are underestimated or inaccurately incorporated into a backtest, they will only show up once their often significant impact on the bottom line has become apparent. Accurate and granular accounting of costs tailored to the particular instrument and the size traded avoids this situation. In order to construct robust investment strategies, it is essential that asset managers accurately incorporate all trading costs including commissions, slippage, bid-ask spreads and market impact into their backtesting process.
Explicit and implicit components
The overall cost of a systematic trading strategy comprises explicit and implicit
components. The explicit component is typically known in advance of trading such as agency commissions and fees. Implicit costs are less observable, harder to estimate and can be of a higher magnitude than the explicit costs.
Implicit costs can be characterised as containing three parts: 01
Implicit cost = Instant impact + Temporary impact + Permanent impact
Instant impact: Cost incurred immediately, such as crossing the bid/offer spread or incurring slippage
Temporary impact: Adverse market price movement during the execution of the trade
Permanent impact: Difference in market price before and after trade
The temporary and permanent impacts are sometimes grouped together as market impact. The remainder of this paper focuses primarily on instant impact, the effect this impact can have on the total return of a systematic strategy and potential ways to mitigate it.
Receive the latest news and insights, directly to your inbox.
Consider the following backtests of typical systematic strategies comparing hypothetical performance with, and without costs02.
1. FX Short-term Value Strategy
This FX systematic strategy holds a long-short basket of FX futures and adjust its positions on the back of short-term fundamental signals. In effect, the strategy rebalances weekly with positions in the basket completely reversing from one rebalance to another. The higher rebalancing frequency has the advantage of increased reactivity but the sheer number of trades creates an annualised cost of 1.5% loss in returns. On a cumulative basis over a decade, the strategy’s total return net of costs (1.6% total) ends up a tiny fraction of its hypothetical gross return with zero costs (17.1% total).
This realistic estimation of trading costs suggests improvements to the
strategy, be it running scenarios for lower turnover versions or including selection penalties for currencies with wider spreads and higher volatilities. Furthermore, backtests of systematic strategies within the FX markets typically use a cut-off point at New York close, which in real-life is a less liquid fixing time03. A robust FX backtesting process would instead allow testing for performance across multiple execution time cuts.
This realistic estimation of trading costs suggests improvements to the strategy, be it running scenarios for lower turnover versions or including selection penalties for currencies with wider spreads and higher volatilities. Furthermore, backtests of systematic strategies within the FX markets typically use a cut-off point at New York close, which in real-life is a less liquid fixing time03. A robust FX backtesting process would instead allow testing for performance across multiple execution time cuts.
2. Global Tactical Asset Allocation Strategy (GTAA)
The GTAA strategy analyzes short-term performance across different liquid markets/asset classes and allocates to the instruments that have outperformed on a risk-adjusted basis. The strategy employs daily rebalancing based on a mean variance optimisation. This strategy demonstrates an even more extreme example of the impact of transaction costs on the total return.
By properly accounting for costs, a convincing backtest becomes a clear loss-making proposition. Though the strategy invests in highly liquid instruments with transaction costs of low single digit basis points, the fast reversal of daily positions erodes up to 6.2% a year (from the no-cost backtest annualised return of 3.9%) and 62% cumulatively over a decade.
The GTAA strategy was improved (as shown in Fig 3) using more thoughtful construction: 1) by slowing the rebalancing through employing a threshold and 2) by increasing the lookback window in the return estimation for the mean-variance optimisation. The longer lookback window smooths the signal and reduces the wild swings in its position weights, thereby making a substantial difference to trading costs (0.5% annualised difference between a no-cost backtest vs including costs).
3. Further Examples
Below are further examples that illustrate the impact of trading costs (including transaction costs and market impact) for a range of systematic strategies across asset classes.
Using SigTech for more realistic strategy construction
As seen in the case of the GTAA strategy, thoughtful design of systematic strategies can help control transaction costs. Estimating this cost depends on several variables, especially as costs can vary across trades depending on size, trade type, instrument characteristics, and venues. In order to have a better grasp of what proportion of hypothetical returns are accessible, portfolio managers need a more granular and accurate view of the costs of strategies down to individual orders.
The ability to incorporate the key attributes of individual trades such as bid-ask spread, volatility, volume and the forecasted market impact into the backtest is critical to understand the projected cost of the strategy. Modelling for trading costs is also valuable for strategies in production to determine the slippage between the backtest model assumptions and the live production environment as well as developing a better understanding of its causes.
Previously, portfolio managers had to individually, and correctly, incorporate the relevant costs into their research and backtesting process. Now, technology can provide the required level of detail and accuracy through out-of-the-box transaction cost functionality which can be calibrated as per the user’s needs and integrated with proprietary TCA models and data. A well-designed platform that facilitates easy and accurate construction of backtests that factor in real trading costs can have a significant effect on the bottom line.
‘The ability to incorporate the key attributes of individual trades such as bid-ask spread, volatility, volume and the forecasted market impact into the backtest is critical’
SigTech platform trading costs
The SigTech platform offers realistic and accurate backtesting that takes a holistic view of trading costs into account. Features of the platform include:
incorporating fixed commissions, fixed/percentage-based spread costs and
the flexibility to specify costs based on specific instrument and trade type
the ability to scale market impact non-linearly by trade sizes and participation rates
calibration of parameters to include proprietary estimates of trade costs based on client’s in-house trade information
the flexibility to integrate proprietary transaction cost models
feedback loop from execution
multiple timestamps for certain markets allowing for granular specification of
trade size depending on time of day or VWAP
modelling of stop-loss strategies
time-variant parameter specification to account for scenarios of market-stress/crowding
Unlike in-house solutions, SigTech eliminates the expensive upfront costs of infrastructure build out and data onboarding so clients can focus on investing from day one.
As the only plug and play trading platform of its kind for quant multi-asset research and investment, SigTech reduces time to market allowing users to focus on alpha generation. Infrastructure build-out and data management are rapid. Strategy development and backtesting are also streamlined to maximise the speed of strategy implementation in both research and production environments.
The first end-to-end platform of its kind, SigTech allows users to focus on signal detection and backtesting, significantly reducing the time and cost of investment strategy implementation and deployment.
This document is not, and should not be construed as financial advice or an invitation to purchase financial products. It is provided for information purposes only and is subject to the terms and conditions of our disclaimer which can be accessed at: https://www.sigtech.com/legal/general-disclaimer