Configuring Drop Alerts That Don't Spam You

Why alerts become noise

A raw drop alert on every price movement will fire hundreds of times per day across a full sports board. Most of those are noise — soft-book adjustments, low-liquidity moves, market-normal fluctuations. The skill is in configuring filters that surface only the signals worth reading.

Bad alerting creates two problems at once. First, it wastes attention by forcing you to check movements that were never meaningful. Second, it trains you to ignore alerts entirely because the signal-to-noise ratio collapses. Once that happens, even the good alerts arrive into a system that has already taught you not to care.

The right way to think about alerts is not "how do I catch everything?" but "how do I see the smallest set of movements that still deserve human review?" That is a much better design question.

Core rule levers

Percentage windows

The drop percentage sets how large a move (measured from the opening price) qualifies as a signal.

Window	Typical use
≥8%	Sharp moves only — major steam triggered by significant sharp money or breaking news
≥5%	Strong signals — the sweet spot for most pre-match alerting
≥3%	Subtle moves — useful near kickoff when small moves carry information
<3%	Mostly noise — rarely actionable on main lines

Start higher than you think you need. Most users benefit from seeing fewer, cleaner moves first. Once you know what a qualified signal looks like in your sports, you can lower thresholds with intent instead of curiosity.

Time windows

The time window sets how fast the drop must happen to qualify.

Window	What it catches
≤1 hour	Fast money — sharp-driven moves or news-driven steam
≤3 hours	Normal pre-match movement — most actionable signals
≤6 hours	Slow drift — less reliable, often noise
Any	Unfiltered — use only for post-match analysis

Minimum and maximum time windows can be combined. A drop that happens over 10 seconds is likely a stale price being corrected, not a signal. A drop over 12 hours is probably just regular market movement. Most actionable signals fall between 10 minutes and 3 hours.

Minimum limit threshold

Set a minimum limit at the reference book (Pinnacle) to filter out low-liquidity markets. A limit below €200 on a major-league match suggests the market is thin or the bookmaker is not confident in the price. The exact threshold depends on the sport and market you track. Why Pinnacle Is the Reference Bookmaker explains why that sharp-book number matters more than a soft-book limit.

A useful starting point: filter out any drop where Pinnacle's limit is below €300 for main markets and below €100 for derivative markets. You are not trying to be perfect. You are trying to remove the weakest signals first.

Market-specific alerts

Different markets behave differently. A fast drop on a correct-score market is less meaningful than the same drop on a moneyline because liquidity is lower and single bets can move the line. Configure separate rules for:

• Main lines (moneyline, handicap, totals): use wider windows and higher thresholds — these reflect genuine market pressure.
• Derivative markets (correct score, exact goals, player props): use tighter windows and lower thresholds — these move differently and require separate calibration.

If you force every market type through one shared rule, your alerts will either be too loose for niche markets or too strict for liquid ones.

Choosing a starting profile

Conservative profile

Use this if you are new to market-reading or only want the cleanest signals.

• Drop %: high threshold
• Time window: shorter
• Minimum limit: higher
• Market scope: main lines only

This will miss some good moves. That is fine. The goal at first is to learn what strong signals look like, not to capture every edge case.

Balanced profile

This is the best default for most serious users. You want enough volume to learn, but not so much that the board becomes exhausting.

• Drop %: medium threshold
• Time window: moderate
• Minimum limit: moderate
• Market scope: main lines plus selected totals

Exploratory profile

Use this only when you are deliberately studying a niche sport or testing new market behaviour. It is a research setup, not a production setup. Expect more false positives.

Three worked configurations

These are illustrative examples. Adjust based on your bankroll, sport preference, and risk tolerance.

Configuration 1: Sharp-money scanner

• Drop percentage: ≥8%
• Time window: ≤3 hours
• Minimum Pinnacle limit: €500
• Markets: main lines only
• Use case: catch only the heaviest steam moves driven by significant sharp action.

Why it works: the rule is selective in all three ways at once. You are demanding a big move, a meaningful time window, and enough market confidence to matter.

Configuration 2: Pre-kickoff edge finder

• Drop percentage: ≥3%
• Time window: ≤1 hour (within 3 hours of kickoff)
• Minimum Pinnacle limit: €200
• Markets: main lines + totals
• Use case: catch subtler moves near kickoff when small shifts can carry information.

Why it works: as kickoff gets closer, smaller moves can matter more because the market is incorporating final information. The tighter time condition prevents the alert stream from filling with slow drift.

Configuration 3: EV-focused

• Drop percentage: any
• Time window: any
• Minimum EV: ≥3%
• Markets: main lines only
• Use case: focus on positive expected value rather than drop size. Useful when you care about mispricing more than movement direction.

Why it works: this is not a drop rule in the classic sense. It is a value filter. It is useful when you already trust the pricing model and want the alert feed to surface edge instead of momentum.

How to keep alerts useful over time

Review misses, not just hits

If a rule never alerts, it may be too strict. If it alerts constantly, it is too loose. The useful review question is: which moves would I have wanted to see, and which ones wasted my time?

Change one lever at a time

Do not simultaneously lower drop %, widen the time window, and add more markets. If the alert quality collapses, you will not know why. Make one change, observe for a few days, then adjust again.

Match alert density to your workflow

If you are checking the board casually between other work, you need fewer alerts than someone actively monitoring a session. Good configuration respects the amount of attention you can actually give the feed.

Alert ≠ instruction

An alert is a notification that a signal exists. It is not a recommendation to bet. Every alert should be read against the operational checklist (opening price, curve, limit, timing, available price) before you decide whether to act. See The Drop Model for the complete framework, and use the Glossary when a term or market type needs a quick refresher.

The discipline here is easy to say and hard to keep. Alerts create urgency by design. Your process has to actively cancel that urgency by forcing a read before a decision.

Example review loop

At the end of each week, take ten recent alerts and sort them into three piles:

1. Worth reading and still actionable.
2. Worth reading but already gone by the time you checked.
3. Not worth reading at all.

That review tells you what to change. If pile 2 is large, you may need tighter time windows or faster notification handling. If pile 3 is large, your filters are too loose. If pile 1 is tiny, you may be too restrictive.

Build the rule in PhotonOdds at Rules Settings once you know your drop %, time window, and minimum-limit thresholds.