Is the momentum premium finally being arbitraged away?
Head of Equity Strategy, Saxo Bank Group
Summary: More and more momentum strategies and momentum-based funds are popping up, but this factor is being arbitraged away much faster than their promoters are willing to admit.
We have been using the momentum factor in our various trading strategies for years. We are sure that almost any quant-driven outfit has various forms of momentum factors featured in their machine learning models. On top of that there is the old equity factor guard that implement the momentum factor more purely, as in Fama-French. Momentum is one of those factors with the longest track record of showing a premium over the market return, but the period following the Great Financial Crisis has been tough for the momentum factor. The duration is now so long that it raises the question of whether the momentum factor has been arbitraged away.
Why is this question so important? Because as I wrote two days ago, the number of momentum-driven strategies is staggering. The marketing machine behind momentum is large and this is likely because it is simple to explain for retail investors and the entry barrier is very low as the trading strategy is not very sophisticated.
We feel an obligation to warn about momentum investing because we believe it will not deliver the same returns as it did before.
If we start with the market-neutral momentum factor, then the 'long-only US momentum portfolio' chart below shows why momentum investing has been so popular. Excluding the pre-WWII period, momentum investing generated 0.8% monthly alpha before trading and financing costs. The big question is whether the momentum after these costs was all that profitable because accessing shares to short-sell was not easy until the 1990s. What stands out, though, is the period after the GFC: we observe a clear dilution of the momentum premium with most months actually showing negative premium. Another important observation is that market-neutral momentum factor returns decline during bear markets. If this behaviour holds going forward, then momentum investors will likely experience poor returns over the coming years as we are likely closer to the end of this bull market.
The degrading in factor premium starts earlier than in the market-neutral but since both portfolios do not include implementation costs it is likely that the degrading starts around the same time. We have plotted key events around momentum investing with the famous 1993 paper by Jegadeesh & Titman that was the first academic paper illuminating the momentum factor. However, some variations of this momentum factor have likely been used by hedge funds during the 1980s as the breakout trend-following systems used by futures and commodity traders broke down.
The two ETF launches on momentum from AQR and iShares in 2009 and 2013 respectively have opened up the gate to the public expanding momentum investing’s footprint on the market. What we observe in the long-only momentum portfolio is again degrading under bear markets as well as the worst period for momentum since the 1930s. The estimated average monthly premium versus the S&P 500 is around 25 basis points, which is still good but keep in mind that this is before implementation costs.
If we do the comparison with the iShares momentum ETF on US equities you will not find alpha either. To be fair, against the momentum factor the recent degrading is within the statistical confidence band and might very well rebound to its previous monthly premium. Cliff Asness, the CEO and co-founder of AQR, recently talked about the disappointment in factor investing in his Liquid Alt Ragnarök paper.
AQR’s momentum fund has $1.08 billion in AUM as of yesterday and the iShares Edge MSCI USA Momentum Factor ETF has $10.02bn as of yesterday, up from $4bn last year in October. On top of that, momentum is used as a feature in machine learning models and is implemented by do-it-yourself investors and discretionary portfolio advisors.
As momentum investing is a high-turnover and liquidity-demanding investing strategy, it means that we may be close to, or have already exceeded, the capacity for momentum investing. As access to computer power grows and implementation costs fall, all simple investing strategies will be arbitraged away. The momentum factor will not escape this fact. Our view is that momentum investing will disappoint going forward and the sharp rise in momentum funds everywhere is a frightening signal.
In Denmark, we are actually seeing momentum strategies being sold to the public on the leading benchmark index OMXC25. Think about this for a while.. the momentum factor is a broad-based premium that has held up over many decades (excluding the recent degrading), but when momentum is chased in a small subset of the total sample it becomes very speculative and likely data snooping – or fooled by randomness, as Nassim Taleb would put it.
Let’s say that you invest in the 10 best-performing stocks in the OMXC25 based on some version of the momentum factor. What you are really betting on, then, is that out of the 5,000 stocks in the MSCI World Index these 10 shares, constantly rebalancedc will deliver consistent alpha. This is despite the fact that the broad-based momentum factor has shown sharp degrading.
Yes, the momentum factor can be implemented in different ways, but betting on a subset of 10 stocks out of 5,000 is to us a very high-risk strategy that could leave one chasing a mirage.
Again, simple strategies will be arbitraged away.
Please read our disclaimers:
- Notification on Non-Independent Investment Research (https://www.home.saxo/legal/niird/notification)
- Full disclaimer (https://www.home.saxo/en-gb/legal/disclaimer/saxo-disclaimer)