Two rules test an upward trend in different ways: buy sustained strength, or wait for a short dip to start recovering. Each has to earn enough to cover trading costs.
The same BTC, ETH and SOL observations feed four separate $1,000 paper accounts. They are alternative scenarios; their balances are never added together.
Reads
Public Kraken order books, sampled every minute; 24 continuous hours are required before entries. Entry rules run every 15 minutes.
Records
Buy, wait and exit decisions, delayed simulated fills, costs and account value.
Slow trend
Looks for
A rise that holds across the past day, six hours and one hour.
Acts when
All six strength and average-price checks below pass, along with data, budget and risk checks.
Why it might fail
A strong move can reverse after entry, or be too small to cover costs.
Trend pullback
Looks for
An established upward trend with a short dip that starts recovering.
Acts when
Daily gain is at least 2%, the six-hour average is above the 24-hour average and price remains above the 24-hour average. The past hour is down 0.25–1.5%, but the last 15 minutes recover at least 0.1%.
Why it might fail
The dip may be the start of a larger fall.
How both leave a position
They request an exit after a 4% price drop from entry, a 6% drop from the observed high, a fall below the 24-hour average, or 72 hours held. These are triggers, not guaranteed sale prices. A $100 entry budget includes fees; there can be at most three positions and a six-hour wait before re-entering an asset.
What they are compared with
Cash makes no trades and assumes no interest. Hold 30% basket buys $100 of each asset once and keeps the remaining cash. Its exposure and exit rules differ, so a raw return comparison is incomplete.
Stale data blocks entries. A 3% daily account loss or 8% fall from peak value persistently halts new entries; restarting does not clear the halt. The full rules and cost assumptions remain part of this study design.
THE HYPOTHESIS
What we’re asking.
If strength persists across several time windows, a rule-based entry may capture some of the continuing move. It must earn enough to cover entering and exiting.
WHY THIS TEST
A manageable question.
We chose three established spot markets to reduce the number of moving parts. Two active rules share the same market observations, with separate cash and holding comparisons.
FROM AN IDEA TO A RECORDED DECISION
How the study works.
01
Observe first
Collect 24 continuous hours of forward order-book observations. Gaps can prevent a decision; old history is not backfilled to rush the start.
02
Require agreement
The Slow trend rule checks daily, six-hour and one-hour strength, plus short and long averages. All entry conditions must pass.
03
Queue, then simulate
A qualifying signal creates a paper order. It waits for a later book and uses displayed depth plus explicit cost assumptions. A signal is not a fill.
04
Watch for a reason to leave
Exits monitor trend reversal, a price stop, a trailing stop and a time limit. Account loss limits can halt entries. The final exit price is not guaranteed.
SLOW TREND / ALL SIX MUST PASS
What the Slow trend model checks.
01
24-hour change
At least +2%
Ask for sustained daily strength.
02
Six-hour average
Above the 24-hour average
Check whether recent prices support the larger direction.
03
Latest price
Above the 24-hour average
Avoid entering below the long reference.
04
Six-hour change
At least +0.5%
Confirm strength beyond a single short interval.
05
Latest price
At or above the six-hour average
Keep the newest observation aligned with the shorter trend.
06
One-hour change
Zero or positive
Wait if the most recent hour is falling.
Fresh data, warmup, available capital, spread, existing positions and cooldown checks also govern whether an order can be queued. These thresholds describe this particular study, not a recommended trading recipe.
WHAT COULD CHANGE THE INTERPRETATION
The limits belong beside the idea.
The model uses a 0.8% taker-fee assumption and 0.05% adverse-price allowance on each side, plus observed spread/depth. These are frozen research assumptions, not a quote for your account.
Each active account starts with $1,000 of paper capital and uses a $100 entry budget. The holding comparison buys $100 of each asset once, leaving about $700 in cash; exposure differs.
A valid test needs at least 30 days after eligibility and a final seven-day confirmation period. A few overlapping trades cannot settle the question.
WHAT WE NEED TO LEARN NEXT
Does the evidence survive closer inspection?
The next question is whether any advantage remains after costs, missed data and comparison with holding or cash. Results will include every completed position and unresolved exit.
Detailed recorded results are not publicly released yet. This page describes the experiment; it does not establish profitability.