Overview
Every Greenscreens rate prediction is accompanied by a Confidence Level. This is an AI-generated score that represents the rate's likelihood of accuracy. When the AI machine learning model has plenty of data to work with and historical numbers are fairly consistent, predictions will generally be more reliable, and you'll see a higher Confidence Level on your rates.
The Confidence Level keeps the prediction process transparent. If the AI doesn't have enough data to narrow down the spread of likely rates, you need to know that, so you can be more cautious in your quoting.
The Confidence Level tells you how likely the rate is to be accurate. It is determined by the amount and quality of information available to the AI. Confidence Levels usually fall between 50 and 100.
Confidence Level Sources
The confidence number is primarily determined by three things:
Density of Historical Data
Confidence Levels tend to be higher on lanes where there is a larger quantity of historical data. The model takes into account both broker-specific data and data provided by the entire Greenscreens.ai network.Market Volatility
The model takes note of market fluctuations. It considers the rate at which truck rates change over time, both in frequency and magnitude. It also takes into account capacity supply and demand conditions in origin and destination market. A more stable market usually means a higher Confidence Level.Spread of Potential Outcomes
When the variance in historical prices is minimal, the model tends to be more confident in its predicted rate. Where prices fluctuate substantially, the model tends to be less confident, as it has a wider band of previous load data to consider.
Confidence Suggestions
Low Confidence (62% and below): We suggest doing one or more of these things:
Get multiple bids from carriers before accepting a price.
Consider starting negotiations with the Start Rate.
Review the Greenscreens Network Rate and compare its Confidence Level with that of the Target Buy Rate. If the Network Rate has a higher Confidence, use it instead of the Target Buy Rate.
Give yourself additional lead time in booking the load.
Review the data shown by the Similar Lanes feature.
Add some additional margin to make sure you’re covered.
Medium Confidence (63% - 75%): We suggest getting multiple bids from carriers before accepting a price or adding some additional margin to be sure you're covered.
High Confidence (76% - 87%) & Very High Confidence (88% - 100%): A high or very high Confidence Level suggests that you can book now at the given rate.
Target Confidence vs Network Confidence
Greenscreens recommends taking both Network and Target Buy Rate Confidence into consideration when deciding how to cover a specific lane.
For every prediction, Greenscreens provides a Network Rate prediction and a Target Buy Rate prediction. The Network Rate prediction is produced by the network model, which uses our entire network's load data. The Target Buy Rate prediction is created by a model trained to predict a brokerage’s rates, based both on the brokerage’s data and data from Greenscreens’ entire network. Since the Network Rate prediction model draws on a different data set and is tuned to predict a different outcome from the Target Buy Rate model, Confidence Levels will vary between these two predictions.
The Network Rate Confidence Level is just like the Confidence Level on the Target Buy Rate, but it's based on data from the entire Greenscreens network and isn't biased toward any one particular brokerage. This can be useful when a lane is new to your brokerage but has seen a lot of business from other brokers in the network. When you see a low Confidence Level on the Target Buy Rate and a high Confidence on the Network Rate, we recommend that you use the Network Rate in your quote, because a higher Confidence indicates that the prediction is better supported by available data.
If a brokerage has a smaller spread of truck rates on a given lane, with more volume consistency, the target model may produce a higher Confidence Level. If a brokerage has no historical volume on a lane (and the model was unable to rely on history from other brokerages that were deemed to be similar or other lanes that would be strongly correlated to the requested lane), where the network has a large quantity of historical volume, the Network Confidence will usually be higher.