EdgeScore
EdgeScore tools

Asset rotation backtest

A rotation rule ranks a pool of instruments by their past return each month and holds the top few until the next month-end. Set the rule, and see how it would have behaved on past closing prices, next to simply holding the whole pool and next to the market index.

Prices: Bucharest Stock Exchange (bvb.ro), end-of-day closes. Past results of a rule say nothing certain about its future results.

Loading instruments…

How the rule is run

  1. At the last trading day of each month, every instrument in the pool gets its return over the lookback: this month-end close against the month-end close that many months earlier.
  2. The pool is ranked by that return and the top N are selected, each at an equal share (1/N).
  3. With an absolute filter, a selected instrument whose return is not above the threshold, or whose close is not above its 200-day average, has its share moved to cash, which earns the cash return you set.
  4. The portfolio trades at that same close and is held to the next month-end; the trading cost is charged on the share of the portfolio that changes.

No look-ahead

Each decision uses only closes dated on or before its decision day: the two month-end closes for the lookback and the 200 daily closes ending that day. Removing every later price leaves all earlier decisions unchanged; our tests check exactly that. The month in progress is never a decision date, so “this month” means the rule applied to the last completed month-end.

One idealisation remains: the rule trades at the same close it measures. In practice an order placed after the close fills at a later price.

Limits

  • Share prices are price returns, without dividends. BET-TR reinvests dividends, so the comparison with it is tilted against the rule and the equal-weight pool; BET (price only) is the like-for-like index.
  • Survivorship: the pool is chosen today from instruments that still trade, which flatters any backtest of them.
  • Few instruments and thin trading on some shares mean a single name can drive the result. Taxes are not modelled.
  • Trying many settings and keeping the best one finds noise as often as a pattern. The “every lookback” table shows how fragile a choice is.

For the current momentum of every instrument, see the momentum table.

For information and education only, not investment advice. Data may contain errors.