Direct answer
Timeframe affects TSI because the indicator is computed from price history over a defined observation window. Changing that window changes which swings are considered “momentum,” how much short-term noise is included, and how quickly the indicator responds to turning points.
Mechanism and definition
TSI is typically built from momentum computed over a chosen set of lookback periods, then smoothed. In plain terms, it answers: “How strong is the recent directional change in price, compared with a baseline?” The key link to timeframe is that “recent” depends on the lookback length.
When you choose a shorter timeframe (shorter lookbacks), the calculations use fewer bars. That makes TSI more sensitive to each new price move, so it can change quickly as soon as momentum shifts. When you choose a longer timeframe (longer lookbacks), more past data is blended into the calculation, so random short-term fluctuations tend to average out.
A simple way to reason about sensitivity is this: the indicator is not observing the future; it is re-expressing past price differences. If the timeframe changes, the set of past differences changes, so the indicator’s path changes too.
Evidence through an example scenario (with assumptions)
Scenario: You run the same TSI settings except you change the lookback length, and you apply it to a sequence of prices that alternates between two phases:
- Phase A: steady upward movement with small pullbacks.
- Phase B: choppy sideways movement with frequent short swings.
Assumptions for the example:
- You use the same price series and the same sampling frequency.
- You apply no changes to costs or data cleaning.
- You compare the indicator shapes rather than treating any specific level as a standalone signal.
What you would usually observe conceptually:
- In Phase A, both short and long timeframes can reflect positive momentum, but the short-timeframe TSI often responds faster to pullbacks.
- In Phase B, the short-timeframe TSI is more likely to fluctuate because it “counts” more of the small reversals as meaningful momentum. The long-timeframe TSI typically changes more slowly because smoothing dilutes those reversals.
Material limitation: this is about indicator behavior, not outcomes. A smoother or faster indicator does not guarantee better results, because real trading performance also depends on spreads, execution, and regime changes.
Limitations and verification (what can fail)
-
Noise and regime dependence: Short timeframes can turn frequent minor moves into large indicator swings, especially in range-bound or low-trend conditions. That can make interpretation unstable.
-
Lag and missed turns: Long timeframes can smooth volatility, but that smoothing can delay changes. In fast reversals, a delayed indicator can describe momentum only after the move has already changed.
-
Sensitivity to assumptions: “Same indicator” can still differ if data frequency, missing data handling, or parameter choices change. Even if the formula is unchanged, the computed values depend on the exact input series.
-
Costs and execution: Any historical pattern you notice might not transfer when adding friction (spread, commissions) and when execution is not instantaneous.
Controlepunt (control point): To verify timeframe effects independently, keep the dataset consistent, change only one timeframe-related parameter at a time, and compare out-of-sample periods. If the behavior only “works” in one historical regime, it is evidence of timeframe-specific fitting rather than a stable property.
Verification or next question
If you want to go further, a useful next question is: “Which timeframe choice best matches the kind of momentum I want to measure—quick swings or broader directional change—and how stable is that behavior across different historical regimes?”