Bet Angel Guardian: The Complete Automation Guide

What Guardian actually is

This sub sits inside the Betfair Automation pillar. Guardian is a visual rules engine that lives inside Bet Angel Professional (£32/month or £499/year — there is no Guardian-only purchase). It watches Betfair market data, evaluates rules you've defined, and when a rule's trigger fires it places, cancels or modifies bets on your behalf.

Guardian sits alongside Bet Angel's other automation tools — the basic stop-loss/take-profit on the ladder, the Tools menu's hedging calculators, and Bet Angel's Excel integration. Guardian is the most flexible of those. It is also the steepest learning curve.

For broader context: the best Betfair bots 2026 ranking places Guardian first overall, and the automation rules and triggers piece explains the model from a tool-agnostic angle.

The trigger and rule model

Every Guardian automation is a set of rules. Each rule has:

• A trigger — the condition that fires the rule. "If the last price traded is below 3.50". "If volume in the last 30 seconds is above £2,000". "If we are 2 minutes from the off". Conditions can be combined with AND/OR.
• An action — what happens when the trigger fires. Place a back order, place a lay order, cancel orders, green up, close all bets in a market.
• An enable/disable state — rules can be on or off. Rule firings can toggle other rules. This is how Guardian builds state machines.
• A scope — applies to every market, only this market, only this selection.

The model is identical in concept to Fairbot's automation, BetTrader's automation, or what you'd write in Python. Skills transfer between tools. The Guardian-specific value is the maturity of the implementation and the size of the community-supplied rule library.

Building your first rule

The simplest useful Guardian rule is "stop loss at 3 ticks". Imagine you've manually placed a £20 back at 3.40 and you want Guardian to automatically lay the position out if the price drifts to 3.45 (3 ticks against you).

1. Open Bet Angel, open Guardian (Tools → Guardian).
2. Right-click in the rules grid, "New Rule".
3. Name it "3-tick stop loss".
4. Conditions tab: add "Last price traded is greater than or equal to 3.45". Scope: "Selection".
5. Actions tab: add "Place bet — LAY at Best back, £20 stake".
6. Once-only checkbox: ticked (so it fires once, not on every poll).
7. Save. The rule now runs on the selection of the market that's currently active.

You've replaced what would otherwise be a manual click. Multiply that across 30 markets a day and Guardian's value is obvious.

Trigger types you'll actually use

Guardian exposes around 30 trigger types. The ones used in 90% of working rules:

• Last Price Traded. The current LTP, in or out of range. The single most common trigger.
• Best back / Best lay. The top of the queue prices. Triggers like "lay at the best lay price".
• Time to market start. "Within 2 minutes of off", "after the off". Mandatory for pre-race rules.
• In-play markers. "Match is in-play", "Game time over X minutes".
• Volume. "Matched volume over £X", "Volume in last N seconds over £X". Used to filter for real liquidity.
• Trader profit/loss. "Selection P&L greater than +£10". The basis of every Guardian-side take-profit and stop-loss.
• Weight of Money. The Bet Angel-calculated imbalance between back and lay queues. Some traders use it; behaviour can be unstable on thin markets.
• Rule states. "Rule X has fired Y times" — for state-machine logic.
• External value. A cell in a linked Excel sheet. Lets you bring in custom data, your own ratings, anything you can compute.

Action types and persistence

Action types you'll use:

• Place bet — BACK / LAY. At a chosen price reference (best back, best lay, LTP, LTP±N ticks) with a chosen size and persistence (Lapse / Persist / SP).
• Cancel bets. All unmatched bets in this market, or by selection.
• Green-up bet. Guardian calculates and places the offsetting bet that levels P&L across all runners — the same maths covered in green up explained.
• Close all bets. Closes the position at market price.
• Enable / disable rule. Toggles another rule on or off. The state-machine building block.
• Send alert. Audio cue or pop-up. Useful when you want notification without execution.

The persistence choice is non-trivial. Lapse (cancel at the off) is right for pre-off-only rules. Persist (keep into in-play) is right for SP-based strategies and lay-the-draw style. SP (match at the official starting price) is right for SP-based value plays. Get this wrong and pre-off rules leak into in-play with disastrous fills.

Strategies Guardian handles cleanly

Guardian was designed by traders who themselves trade Betfair, so the catalogue of cleanly-expressible strategies is large. These are the ones we've seen run reliably:

• Lay the Draw. Lay the draw at the off, exit on goal with green-up, exit at 80 minutes if still 0-0. Walkthrough in our swing trading and lay the draw guide.
• Pre-race scalping. Enter on momentum + volume gates, exit at LTP+1 tick with stop at LTP-2 ticks. Covered in scalping on Betfair.
• Dobbing (double-or-bust). Back at SP, lay at half-price in-running. Specific to horse racing.
• Lay favourites. Lay drifting favourites with volume confirmation.
• Tick-based stop-loss/take-profit. After a manual entry, Guardian closes the position at +N or -M ticks.
• Pre-match price drift backing. Back selections that have drifted N ticks in the last hour pre-off.
• In-play swing trading. Enter on price extremes, exit at mean reversion.
• SP value plays. Place bets to match at SP when an Excel-loaded rating indicates value.

Where Guardian hits its limits

• Custom data integration. Excel is the only external data hook. If you need a Python model or REST API in the loop, Guardian can't host it.
• Heavy multi-market parallelism. Guardian works comfortably on 10–15 markets simultaneously. Beyond 25 it slows materially and rule firings can lag.
• Complex state. Three or four chained rules is fine; ten chained rules becomes painful to debug.
• ML in the loop. No native way to call a Python or REST model on every tick. You can fake it via Excel + a separate Python process writing into Excel, but it's brittle.
• Speed-critical scalping. Guardian's rule polling adds ~150–250ms vs. a tight Python loop. For 1-tick scalping that matters.

When you hit any of these walls, the answer is usually to add a Python bot for the constrained strategy and leave Guardian running the strategies it does well. The Python tutorial covers the build.

Common mistakes

• No "once-only" on entry rules. The rule fires every poll, you end up with 8 back bets at slightly different prices on the same selection.
• Wrong persistence. Pre-off rule with "Persist" persistence — your unmatched lay sits there into in-play and matches at an awful price.
• No volume gate. Triggers fire on illiquid markets, you get partial fills and bad average prices.
• Stake variable, not fixed. Guardian lets you size by "% of bank" — quick way to accidentally place £200 bets when you meant £20. Use fixed stakes until you trust the logic.
• Running on a Windows laptop that sleeps. Same problem as any bot. Use a dedicated machine or a Windows VPS.
• Not version-controlling rule files. The .bagf Guardian rule files are XML. Keep them in a git repo so you can roll back a broken edit.

Combining Guardian with Python

The mature retail bot trader typically runs both: Guardian handles the well-defined visual strategies (LTD, scalping, dobbing) and a separate Python bot handles strategies that need custom data or ML.

Two integration patterns:

1. Excel as the bridge. Python writes a "signal" value into a cell; Guardian reads the cell as an "external value" trigger. Works but is fragile — Excel hiccups cause both to misbehave.
2. Separate accounts or separate markets. Guardian on one set of markets (or one Betfair account), Python on another. Cleaner; avoids any shared-state issues.

If you go all-in on Python, see profitable bot strategies for setups that don't require Guardian's UI.

FAQ

Can I use Guardian without Bet Angel Professional?

No. Guardian is included with Bet Angel Pro and there's no standalone version. Bet Angel Basic doesn't include it.

What does Guardian cost?

£32/month, £180/quarter or £499/year for Bet Angel Pro. No additional fee for Guardian.

Is there a free trial?

Bet Angel offers a 14-day free trial of Pro that includes Guardian. Long enough to test a simple strategy.

How many rules can I run at once?

No hard limit, but performance degrades past 30–40 active rules on a single PC. For more, use multiple Bet Angel instances on a VPS.

Can Guardian run on a Mac?

Only via Parallels, Wine or a Windows VPS. Bet Angel is Windows-only natively.

Where can I find shared Guardian rule files?

The Bet Angel community forum has a "Files" section with thousands of .bagf rule sets. Quality varies; always paper-trade a downloaded rule before going live.

Your first 30 days with Guardian

A sensible learning trajectory rather than diving straight into complex rule chains:

• Days 1–7. Manual trading in Bet Angel only. Learn the ladder, the one-click betting, the trading P&L view. No Guardian yet.
• Days 8–14. Write your first single-rule automation: a stop-loss that lays out a manual back at -3 ticks. Run it on three races at £2 stakes.
• Days 15–21. Add a take-profit rule alongside the stop-loss. Two-rule state machine. Watch how the rules interact when both fire close together.
• Days 22–30. Pick one of the strategies in profitable bot strategies and build it as a three-rule chain. Paper-trade it for the rest of the month.

The temptation is to skip straight to a 7-rule strategy. Don't. The skills compound; the first three rules teach you 80% of what you need to know about Guardian's state model.

A full walkthrough: lay-the-draw with auto-green

To show Guardian end-to-end, here's a working lay-the-draw rule set against a typical Bundesliga match. Pre-match lay the draw at the available price up to 4.20; if a goal goes in after the 5th minute, green up; if minute 80 arrives at 0-0, close at market.

The state machine: WATCHING (only Rule 1 enabled) → entry placed → IN_TRADE (Rules 2 and 3 enabled) → exit fires → CLOSED (all rules disabled). For the next match, re-enable Rule 1 manually or with a Rule 4 that watches for "market closed = true" and re-arms.

The first time you build this, allow 6–8 hours of trial-and-error. The hard bits are the gating conditions (forgetting the liquidity gate; forgetting LAPSE persistence; forgetting that "score = 0-0" only fires once per goal-state transition). The full match-level strategy is in lay the draw guide.

Advanced Guardian techniques

Three techniques that separate hobbyist Guardian users from serious ones:

• Excel feedback loops. Use Guardian's "set Excel cell" action to write trade decisions to an Excel sheet, then use Excel formulas to compute statistics (running P&L, win rate, average tick movement) and read those statistics back as gating conditions. You build a self-tuning rule set that throttles down during bad runs.
• Conditional stake sizing. Stake size pulled from an Excel cell that's computed as a function of bankroll, edge and current drawdown. Bet Angel calls this "Excel-driven stake". Implements Kelly-style sizing without writing code.
• Cross-market state. Use a master Excel workbook that's read by multiple Guardian instances on multiple Betfair accounts. Coordinate bankroll allocation across markets without any rule knowing about the others directly.

These techniques are the reason Guardian remains relevant even compared to Python bots — once you wire in Excel, you have an extensible automation surface that most retail traders never need to outgrow.

Is Guardian cost-effective at retail stakes?

Bet Angel Pro is £32/month (£384/year). At £20 stakes the bot needs to generate roughly £32 net per month just to pay the subscription. Most beginning bot traders break even on the subscription within 60 days; many never break even because the strategy never finds an edge. The honest framing:

• Stakes £5–£20. Guardian is loss-leading — subscription costs eat the small wins. Worth running anyway as a learning investment; don't expect net profit.
• Stakes £20–£50. Guardian is break-even to mildly profitable. The trader is paying tuition through the subscription.
• Stakes £50–£20