The Honest Record

How our Harness calls actually performed · 206 settled races

Most tipping sites show you their winners and quietly forget the rest. We do the opposite: here's every call, scored against the result — including where backing blindly loses money. We'd rather you trust us than be impressed by us.

37.9%

Market fav win

72.8%

Market fav place

-25.0%

Market fav flat ROI

36.3%

Our pick win

71.7%

Our pick place

-31.7%

Our pick flat ROI

"Market fav" = blindly backing the shortest-priced runner (206 races). "Our pick" = the runner our model rated #1, where it had rated the race (113 races and growing as more races are rated).

By race shape

Our call changes with the race. This shows whether each type of call held up — and the stance we give for it.

Dominant favourite thin we say back to win

Market fav: 88 races Win 48.9% Place 83.0% Flat ROI -25.1%

Our model pick: 54 races Win 50.0% Place 83.3% Flat ROI -24.3%

Competitive thin we say win or place

Market fav: 50 races Win 40.0% Place 70.0% Flat ROI -10.7%

Our model pick: 23 races Win 34.8% Place 65.2% Flat ROI -24.8%

Weak favourite noise we say place or pass

Market fav: 29 races Win 27.6% Place 69.0% Flat ROI -25.0%

Our model pick: 18 races Win 16.7% Place 66.7% Flat ROI -54.2%

Open race noise we say place, value, or pass

Market fav: 39 races Win 17.9% Place 56.4% Flat ROI -42.8%

Our model pick: 18 races Win 16.7% Place 50.0% Flat ROI -40.6%

Read it like this: on Dominant favourite races we say back to win, and those win most often. On Weak favourite and Open races we say don't back the favourite to win — so the low win rate there is the call working, not failing: we steered you to place or pass.

Model vs market accuracy. Across 1,042 rated runners, our model's probability error (Brier 0.0792) is a fraction behind the market's (0.0785). We don't beat the market on raw prediction — almost nobody does — which is exactly why our calls lean on race structure and discipline rather than claiming to out-pick the price.

Our pledge. We will never hide a losing run, cherry-pick a hot week, or claim an edge the numbers don't show. This page updates automatically from real results — good and bad. If a call type stops working, you'll see it here first.

Sample tags: noise under 40 · thin 40+ · fair 120+ · solid 300+ races.

Calls reconstructed from settled results using the same engine the site runs live. Flat ROI = level-stake (1 unit) win bet, scored on settled results. "Market fav" backs the shortest-priced runner every race; "Our pick" backs our model's top-rated runner where the race was rated. Past performance is not a guarantee. Not betting advice · 18+ · GambleAware.