3.95%
best MAPE on held-out data
Property valuation, validated on a temporal holdout split.
The prediction platform
where trust meets insight.
Finally — anyone can get forecasts grounded in the facts of their business. Property prices, freight quotes, crop yields, solar output and more, each with a verified confidence interval — through one REST API.
Enter your website and we'll identify prediction opportunities you could deploy today.
dryapi.app — Prediction Studio API online
Predict
Charts
Share
Publish app
POST /predict — request
{
"model": "property/valuation/v1",
"features": {
"sqft": 1850,
"beds": 3,
"baths": 2,
"zip": 90210
}
} Sale price forecast — 90% confidence band
{
"value": 615800,
"interval": [560400, 671200],
"confidence_interval": {
"lower": 560400, "upper": 671200, "coverage": 0.9
},
"latency_ms": 4.5,
"top_factors": [
{ "feature": "sqft", "impact": "+73.1%" }
{ "feature": "beds", "impact": "+21.4%" }
{ "feature": "location", "impact": "+5.5%" }
]
}
Trusted by sectors that run on forecasts
17
Prediction APIs
9
Industries
0.9
Calibrated coverage
<10ms
Median latency
Connect your data, then predict on it
Stripe Shopify HubSpot Salesforce Google Sheets GitHub Notion Slack Airtable Pipedrive
Every prediction ships with a confidence band.
A point forecast is a guess with extra steps. DryAPI returns a verified interval — a 90% confidence band measured on held-out data, not an approximation.
Every forecast is calibrated on held-out data before it ships, and re-checked as new data arrives — the number you get is the number you can act on.
Why calibrated
response — property/valuation/v1
{
"model": "property/valuation/v1",
"value": 512400,
"interval": [468900, 559200],
"confidence_interval": {
"lower": 468900,
"upper": 559200,
"coverage": 0.9
},
"confidence": { "level": 0.9, "score": 0.93 },
"latency_ms": 7,
"top_factors": [
{ "feature": "sqft", "impact": 0.42 },
{ "feature": "zip", "impact": 0.31 },
{ "feature": "beds", "impact": 0.18 }
]
} Self-serve SIGNAL · SLACK/DISC
Predictions in the tools your team already uses.
Drop a question into Slack or Discord, get a forecast with the confidence band attached — no dashboard tour required.
D dryapi · predictions
J
Julia 4:02 PM@DryAPI can you forecast next quarter's freight costs for the west-coast
corridor?
D
DryAPI 4:02 PMSure — a 90% interval for LA → Seattle, based on 18,400 historical shipments.
freight/quote/v1 — quarterly forecast
$5,180
point forecast
$4,900 – $5,460
90% interval
0.91
calibration
Q6_west_coast_forecast.json
D #forecasts
M
maya Today at 4:02 PM@DryAPI what's the LA → Seattle freight forecast for Q6?
D
DryAPI APP Today at 4:02 PM90% interval on 18,400 shipments — $4,900 to $5,460.
freight/quote/v1 — quarterly forecast
$5,180
point forecast
$4,900 – $5,460
90% interval
0.91
calibration
Q6_west_coast_forecast.json
Benchmarks MAPE 3.95%
3.95%
Best MAPE
Property Valuation
0.988
Best R²
Property Valuation
7ms
Avg response
Across all APIs
90%
Interval coverage
Verified out-of-sample
Calibration curve
out-of-sample
Numbers we can prove on held-out data — not marketing slides.
Calibration is a contract. Every published metric is measured on data the models never saw during training.
Always fresh RETRAIN · DAILY
Models retrain daily. Drift never gets a vote.
New data flows in every night, every model in the catalog refreshes each morning, and drift is re-measured against the latest holdout — so the interval you act on stays calibrated.
- 00:00 Ingest 48 feeds listings · permits · sales
- 06:00 Retrain 17 models refit + recalibrate
- 06:12 Hot-swap registry zero-downtime serve
Production-ready UPTIME 99.9%
The guarantees your forecasts run on.
Uptime, latency, and support — the numbers behind the number. Your forecasts ship on an API that's always warm and always monitored.
99.9%
Uptime
Monitored around the clock
7ms
Avg response
No queues, no warm-ups
24/7
Support
Email + priority SLA
0 ops
For you
We run the models
The proof HOLDOUT 90%
Numbers we can stand behind.
Every model is measured on data it has never seen, recalibrated daily, and shipped with the confidence band attached — no marketing slides.
Property Freight Solar AgTech Commerce Logistics
90%
interval coverage out-of-sample
Every prediction ships with a calibrated confidence band — you see the uncertainty, not just a number.
Daily
retrain cadence across all 17 models
Models are re-fit on fresh data every day and hot-swapped into the registry to keep drift in check.
The catalog 17 ENDPOINTS
17 endpoints. One contract.
Every endpoint below returns the same shape — a point forecast, a calibrated interval and the top factors behind it — through one REST API and one API key.
Property
4property/valuation/v1Estimate a home's market value from its characteristics and location.property/rental/v1Predict fair monthly rent for a property in any market.property/risk/v1Assess flood, fire, and disaster risk for a property.property/investment-score/v1Score a property's attractiveness as an investment.
Construction
3Agriculture
3Energy
2Vehicle
1Freight
1Delivery
1Shipping
1Marketing
1 Integration SDK ×4
Ship in your language.
First-class SDKs and a single REST contract across the entire catalog. Copy the call, paste your features, get a calibrated answer.
predict.py
from dryapi import DryAPI
api = DryAPI("pk_live_...")
res = api.predict(
model="property/valuation/v1",
features={"sqft": 1850, "beds": 3, "baths": 2, "zip": 90210},
)
print(res.value) # 512400
print(res.interval) # [468900, 559200]
print(res.confidence_interval) # {lower: 468900, upper: 559200, coverage: 0.9}
print(res.top_factors) # [sqft +42%, zip +31%, beds +18%]Getting started is easy.
One account, one credit balance, one POST. Pick an API, pay as you go, and get a calibrated answer in milliseconds.
FAQ DOCS →
Questions, answered.
Can't find your answer here? Get in touch.
Tabular forecasting across nine domains — property values and rents, vehicle prices, freight quotes, delivery ETAs, solar output, crop yields and more. Every endpoint returns a point forecast, a calibrated confidence interval and the top contributing factors.