Six weeks of history to judge a 27.75% weekly fall
Should the fund act fully on the downtrend signal at this week's rebalance, or cut the position because the signal rests on only six weeks of data?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 26 March 2018, ETH fell 27.75%, from 523.95 to 378.58, closing 12.7% of the way up from its low and 67.34% below its 52-week high. The prior four weeks returned 2.99%, -16.82%, -25.52% and -2.54%. The worst day was -13.91% on 29 March, the best only 2.23% on 30 March; the largest one-hour move was 6.49% on 30 March at 05:00 UTC. Volume of 0.51bn was 0.99 times the prior twelve-week average. ETH lost 10.17% against BTC, with 0.93 hourly correlation. A trend fund rebalances weekly but has six weeks of history, so no return percentile exists.
Judgment
The trend signal points one way and the fund should rebalance towards it on schedule, since three of the prior four weeks were down, this week extended the fall, and the close sat near the week's low. But the position must be sized for a signal with six weeks of data and no percentile to compare this week against. Use the 139.6% annualised realised volatility and the 6.49% hour to cap size, and accept that the fall may be an early sample that mean-reverts. The 0.93 correlation with BTC says the fund is not being paid extra for an outright market bet.
What would change it
If ETH recovers above the week's open of 523.95 on a rebalance date, the downtrend signal weakens and the position can be reduced. If the history lengthens enough for this week's return to sit far down the distribution of comparable weeks, size can be raised.
What sets this case apart
The decision turns on how much weight a trend signal deserves with six weeks of data, not on whether the trend is down.