The almanac answers one question, where should I fish today, by reading public instruments and turning them into a decision. It never promises fish. It tells you where conditions favor them, when, and with what rod, and it tells you how sure it is.
Every number carries a small mark that says where it came from. Measured means an instrument on that very stretch of river. Proxied means the nearest gauge on the same river, with the distance stated and anything in between, a dam, a tributary, named. Estimated means we modeled it, and we say how. Forecast is the Weather Service. Statistics are the USGS record for the date. Where we have nothing, the page says "not available" rather than inventing a value.
Rivers without a thermometer. The Boulder and the Shields have USGS flow gauges but no temperature sensor. For them, water temperature is estimated: a seasonal curve fitted to decades of USGS field measurements at the river's own gauge (293 readings on the Boulder at Big Timber, 1965 to 2013) gives the usual midday temperature for the date, and today's anomaly on the nearest rivers with thermometers, the Yellowstone at Livingston and the Stillwater or the Yellowstone at Corwin Springs, moves it up or down. The number is marked estimated, the page says it is good to about four degrees either way, confidence is graded down for it, and "typical for this week" on those rivers comes from the same curve rather than from USGS statistics.
Water clarity is the honest example: no reach we cover has a turbidity sensor, so clarity is an estimate from how fast the river is rising, how far above normal it sits and whether it rained, and it is labeled that way everywhere it appears.
The Fishing Score runs 0 to 100 for each reach and each hour. It is a weighted blend of the things that are known to move trout feeding: water temperature first, then flow against the normal for the date, the direction the flow is moving, estimated clarity, the hour of the day, light, wind, the trend in water temperature, and a little for the barometer. Three technique scores, dry fly, nymph and streamer, use the same inputs with different weights, because a streamer likes a little color and a dry fly does not.
| Factor | Share of the overall score | |
|---|---|---|
| Water temperature | 36% | |
| Flow against the normal for the date | 14% | |
| Hour of the day | 10% | |
| Flow trend over 24 hours | 9% | |
| Clarity (estimated) | 8% | |
| Light | 6% | |
| Wind | 6% | |
| Water temperature trend | 5% | |
| Afternoon heat (air over warm water) | 4% | |
| Barometer | 2% |
Above the blend sit hard limits a nice cloud deck cannot average away: water at 68 F caps every score at 45, 70 F at 25, and 73 F, FWP's own threshold, at 8. Reaches where cutthroat are a large share of the catch use a colder curve, because cutthroat give out about five degrees sooner than rainbows.
Because nine factors that are each merely fine average out near 100, the blend is stretched so that only genuinely aligned hours reach the 90s. Every reach's page shows the breakdown for its best hour: each factor's own 0 to 100, its verdict, and its weight, so any number can be checked against the water. Excellent starts at 88, Very Good at 74, Good at 58. A perfectly aligned day lands in the low 90s; an ordinary good day in the 70s.
Typical for this week. Next to the score sits what the reach usually scores at this time of year. It is not the score of a perfect median day, which rarely happens; it is the expected best hour across a spread of ordinary days for the date, built from the gauge's own record: water temperature between its 25th and 75th percentiles for the date, flow between 70 and 140 percent of the median, small rises and falls, clear to overcast, calm to breezy. A 76 that is eight points better than usual is a better day than a 76 that is eight points worse.
The outlook is discounted. For tomorrow and beyond, water temperature and flow are today's readings carried forward, because we do not forecast them yet. Those days' scores are pulled toward the middle, ten percent per day of lead, so a pleasant-looking Sunday cannot outscore a measured today on an assumption. Each outlook day says so.
The score is version 0.2. Its weights and curves were set from the fisheries literature and from how experienced anglers read water, then tightened in September 2026 so that ordinary days no longer read as exceptional ones. It has not been fitted to catch records. It says so in the small print on every page, and it will stay labeled v0 until enough anonymous reports have come back from anglers to calibrate it against what actually happened.
Confidence is computed from the sources behind each reach, not declared. A reach with its own thermometer and gauge can grade high; a reach borrowing a gauge from twenty miles away, with a dam in between, cannot.
FWP's restriction map is read every refresh and sits above the score. A closure removes the score. A hoot-owl restriction blanks out the afternoon hours so no "best window" can land where fishing is not allowed. When measured water is hot enough to stress trout, the page says so plainly, and says whether that is a rule or only our caution.
Which fish live in a reach comes from FWP's electrofishing surveys, filtered to population estimates and matched to the river mile. The page shows the share of the sample, the typical and largest fish measured, and the year of the last survey. That is what lives there, which is not the same as what you will catch. The notes on what each species is doing this month are natural history from FWP and Montana State University sources, general to the species, never a location.
U.S. Geological Survey National Water Information System · NOAA / National Weather Service API · Montana Fish, Wildlife & Parks fishViewer and Fishing Access Site services · USGS 3D Elevation Program · fish plates by Duane Raver for the U.S. Fish and Wildlife Service and from the U.S. Fish Commission's 1889 Yellowstone survey, all public domain.