- The DOE peg is the EIA weekly retail on-highway diesel price, applied after linehaul as a deterministic contract rule.
- A linehaul model must never learn the surcharge, or billing adds the peg twice.
- On a multi-leg move, apply the peg once to the billed miles, then split that one amount by leg miles.
- Effective date, rounding order, and schedule type are stored on the version. A forecast never switches them.
- Neural hedging is only a warning for a quarterly table review, owned by finance and the contract owner.
Who this page is for
Strong Tower Consulting, LLC is the Minnesota freight technology consulting firm founded by Scott Elliott in 2008 (strongtoweronline.com). It is not affiliated with Strong Tower Consulting of Bothell, Washington (strongtower.consulting or strongtower.ai).
Billing and pricing leads who already maintain customer fuel tables are the audience. The peg adjustment is a deterministic contract rule applied after linehaul. A neural model must not learn the surcharge away. Neural hedging, as this page uses the words, is a feasibility warning only: a forecast of regional basis may inform a quarterly peg-table review. It must not replace the contract schedule on a live invoice. Treasury decisions about fuel purchases sit outside this design, and this page does not describe trades.
Peg anatomy
The EIA weekly retail on-highway diesel price is what contracts call the DOE peg. Unless the contract names a regional series, the national weekly retail figure is the input. It is not a terminal rack price, not a futures settlement, and not a forecast of next week. The table that turns the peg into money has three parts: a base price, an increment, and a charge that is either cents per mile or a percent of linehaul. Those three parts are the schedule. They are not coefficients for a model to rediscover.
The arithmetic is ordinary and should stay visible. Subtract the base from the peg. Divide by the increment. Apply the rounding rule the contract states, which in the worked example below keeps only whole increments. Multiply by the cents per mile, or by the percent of linehaul. Then extend by the miles the contract actually bills, or by the linehaul, and round money once at the end. The result is the fuel surcharge. It is added after linehaul. It is not folded into the linehaul rate.
- Subtract the base from the peg. The gap in dollars per gallon.
- Divide by the increment. Apply the contract rounding rule, such as whole increments only.
- Multiply by the charge. Cents per mile, or the percent of linehaul.
- Extend and round once. By the billed miles or the linehaul. The result is added after linehaul.
A linehaul model that is allowed to see historical all-in revenue will learn the surcharge and then sit under a billing step that adds the peg again. That double count is a design defect, not a fuel event. The companion on margin guardrails keeps the peg outside the rate suggestion for the same reason. This page is the schedule side of that split. The win-probability page at lane bid elasticity should not take the peg in as a feature either.
Illustrative worked peg
The following schedule is illustrative. The peg print of $3.90 per gallon is a teaching input, not an EIA release, not a live quote, and not a regional average. The base is $2.40 per gallon. The increment is $0.06 per gallon. Each whole increment adds $0.01 per mile. The contract bills 400 practical miles. Gap is $3.90 minus $2.40, which is $1.50. Whole increments are $1.50 divided by $0.06, which is 25. Cents per mile are 25 times $0.01, which is $0.25. Extended surcharge is 400 times $0.25, which is $100.00. That $100.00 is applied once.
A second teaching input shows why rounding has to be in the table before anyone argues about a penny. A peg of $3.92 produces a gap of $1.52. Divided by $0.06 that is 25.333 increments. The rule in this example keeps whole increments only, so the count stays 25 and the surcharge stays $100.00. Using the fraction instead would extend to about $101.33 after a half-up rounding of the dollars. Both procedures are deterministic. They are not the same procedure. The contract picks one. The invoice job does not "pick the closer."
($3.90 - $2.40) / $0.06 = 25 whole increments
25 x $0.01 = $0.25 per mile
400 miles x $0.25 = $100.00, applied once
Teaching inputs only. The $3.90 peg print is not an EIA release.
| Step | Figure | How it is obtained |
|---|---|---|
| Peg print | $3.90 per gallon | Teaching input. Not this week's publication. |
| Base price | $2.40 per gallon | Contract base. |
| Increment | $0.06 per gallon | Contract step. |
| Whole increments | 25 | $1.50 divided by $0.06, whole steps only. |
| Cents per mile | $0.25 | 25 times $0.01. |
| Miles billed | 400 | Practical miles named by the contract. |
| Fuel surcharge | $100.00 | 400 times $0.25, applied once. |
Percent-of-linehaul is the other shape, and it replaces the cents-per-mile shape unless the contract explicitly stacks them. Illustrative only: 0.5 percent of linehaul per whole increment, 25 increments, is 12.5 percent. On an illustrative linehaul of $1,600.00 the surcharge is $200.00. You do not also add the $100.00 from the cents schedule. Stacking is a different contract. Silence is not stacking. The rate matrix should store the schedule id on the confirmation so billing cannot guess which shape applied.
Why one national peg mis-charges a multi-stop
Retail diesel is not the same price in every region. The national weekly series is an average. A relay that starts where retail diesel is cheaper and finishes where it is dearer still burns fuel at local prices, while a national peg bills one cents-per-mile figure for every mile. That gap is real enough to matter on a long multi-stop. It is not, by itself, a reason for the invoice job to swap series. If the contract names the national peg, the national peg is what the invoice uses, including on the expensive leg.
Teaching contrast only, not a published basis: the contract peg is the $3.90 national input above, and a regional series the contract does not name is $4.40. The extra $0.50 is about eight increments at $0.06, roughly eight cents per mile, or about $32 on 400 miles if someone substituted the regional series. That substitution would change the rate. It is not a correction. The mis-charge, if finance considers it one, is a reason to renegotiate the table at the next review. It is not a reason to "fix" Tuesday's invoice with a forecast or with a series the contract did not name.
A multi-stop makes the temptation worse because the miles are visible by state and the fuel card is visible by city. The architecture still separates evidence from the billing rule. Fuel-card detail can support a quarterly conversation about whether a regional schedule should be offered. It does not override the schedule id on the confirmation. A stop list is not a new contract.
Allocate FSC across legs without double counting
Apply the peg once to the miles the contract bills, then allocate. On the illustrative through move, the customer is invoiced $100.00 one time. An internal relay splits that same $100.00 by miles. The origin-to-relay leg is 250 miles, which is 250 divided by 400, times $100.00, which is $62.50. The relay-to-destination leg is 150 miles, which is $37.50. The parts add to $100.00. Neither division invoices the customer for $100.00 on its own leg. That would bill $200.00 for one physical trip.
Two contracts are a different case, and they must not be stacked on top of a through charge. If each leg is its own confirmation, each leg runs the peg on that leg's miles and that leg's schedule. There is no remainder to allocate, and there is no third surcharge for the "whole lane." Intercompany settlement between divisions uses the same rule: the miles are charged once in the customer's world. Eliminating the duplicate inside the books is an accounting step, not a second peg. Transportation accounting integration is where that elimination should be specified so a relay cannot collect the surcharge twice.
| Leg | Miles | Share of billed miles | Fuel surcharge |
|---|---|---|---|
| Origin to relay | 250 | 250 of 400 | $62.50 internal |
| Relay to destination | 150 | 150 of 400 | $37.50 internal |
| Customer invoice | 400 | All billed miles | $100.00 once |
Empty miles take a surcharge only when the contract says so. The default in this framework is the loaded practical miles the confirmation names. A dispatch plan that adds deadhead does not create a fuel line. A later mile revision is a new rating event under the same schedule version, not a quiet extra. If the revised miles change the allocation, both the customer total and the leg shares are recomputed from the new miles. One side is not left on the old share.
Schedule types
Every customer schedule id should declare its type before the first invoice. Cents per mile, percent of linehaul, a named regional series, and a flat amount inside a peg band are different contracts. They can share a peg publication and still produce different money. The rating job looks up the type. It does not infer the type from last month's invoice, and it does not let a model choose the type that "fits" the fuel card.
A regional schedule is valid only when the contract names the regional series. Pointing a national schedule at a regional feed because the basis looks wide is a table change, and table changes follow the roles in the later section. A flat band, if a contract uses one, is still a lookup: the peg falls in a band, the band has an amount, the amount is billed. The band edges are data. A forecast of where the peg might be next month does not slide the edge.
| Type | What is calculated | What it must not do |
|---|---|---|
| Cents per mile | Whole increments times cents per mile times billed miles | Must not be applied twice to the same miles. |
| Percent of linehaul | Whole increments times the percent times linehaul | Must not also add cents per mile unless the contract stacks them. |
| Named regional series | The same arithmetic on the series the contract names | A national contract does not silently switch series. |
| Flat amount by band | The dollar amount for the band that contains the peg | A forecast does not move the band edge. |
Store the schedule id, the type, the base, the increment, the cents or the percent, the series name, the rounding rule, and the effective window as columns, not as a note on the customer. SQL data mining is the right place to define that extract so an auditor can replay $100.00 from $3.90 without asking a model what fuel "should" have been.
Effective date and rounding
The publication date and the effective date are different. An illustrative rule, not a claim about any real week, is that a peg stored from a Monday publication becomes effective the following Monday and stays in force for seven days. Using the calendar only as a boundary: a publication date of Monday, October 19, 2026, would be effective Monday, October 26, 2026, through Sunday, November 1, 2026. A pickup on Saturday, October 24, uses the prior version. A pickup on Monday, October 26, uses the new version. No price in this paragraph is an EIA print. The dates exist so the boundary is visible.
A multi-day relay needs one answer when it crosses that boundary. The teaching default is that the pickup date governs the entire move, so the October 24 start keeps one peg on every leg even if the second leg runs on October 26. The other lawful choice is a contract that rates each leg on that leg's start date. Either rule is encodable. Using whichever week's peg is higher is not a rule. It is a bias. Encode one, store it on the schedule, and rate the relay from that sentence.
Rounding order is part of the same version. In the worked example the order is: whole increments only, cents per mile to the cent, extend by miles, and the extension is already on the cent. If a schedule produces a fraction of a cent, round the extended dollars half up to the cent once. Do not round the per-mile figure to the cent, extend, and round again, unless the contract says that double rounding in those words. A one-cent argument on the audit queue is often this order, not a fuel event. The penny tolerance in the invoice audit workflow is applied after this rule, not instead of it.
Audit the invoice against the peg
The expected fuel amount is the schedule output: peg version, schedule id, miles, linehaul if the type is a percent, and the rounding rule. The audit compares the billed fuel line with that expected amount. It does not compare the invoice with a neural forecast, a fuel-card total, or a regional series the contract did not name. A forecast that says diesel "should" have been cheaper is not a discrepancy. A billed line of $100.02 against an expected $100.00 is a one-cent difference and follows the penny rule finance adopted on the audit page. A billed line of $132.00, because someone substituted a regional series, is a real break.
Replay has to be possible months later. Store the peg value that was used, not a pointer that now resolves to a newer week. Store the schedule version. Store the miles. A restated EIA number, if an agency ever restates one, does not rewrite closed invoices. The version on the invoice remains the version that was effective on the rating date. Business process optimization should say who may open a closed invoice, and the answer should not be the rating job.
Missing inputs fail closed. No schedule id, no peg for that effective window, no miles, or two schedule types on one confirmation all stop the fuel line. The job does not estimate a surcharge from recent invoices on the lane. Silence is a queue, the same way a missing proof stops a lumper. The biller sees the confirmation, the schedule version, and the empty peg cell side by side. They do not see a suggested "fair" fuel number from a model.
Who may change the table
Two named roles change the table together. Finance owns the base, the increment, the cents or the percent, the rounding order, and the series the schedule is allowed to read. The contract owner owns which customer confirmations point at which schedule id. Neither role is a model, a night batch, or a clerk clearing one invoice. A change carries an effective date. It does not rewrite invoices already rated. An emergency change is still a version, still dual-controlled, and still prospective.
A quarterly review is the legitimate place for a regional-basis conversation. Finance can look at how far a named regional series sat from the national peg over the quarter, using published history, and decide whether to offer a different schedule at renewal. That decision becomes a new version with a future effective date, or it becomes no change. The review memo is not an invoice edit. Customers who stay on the national schedule stay on it until a confirmation says otherwise.
Access to the table should be boring and logged. Who changed the base, the prior value, the new value, and the effective date are the audit. A spreadsheet emailed over the top of the versioned table is a second schedule, and the rating job must not read it. IT strategy planning is where that single path is owned, so a helpful integration cannot update the peg from a forecast file.
Neural hedging is a feasibility warning
The forecast does not replace the contract schedule on a live invoice. It does not choose the peg, the increment, or the cents.
Neural hedging, on this page, means a forecast of the gap between a regional diesel series and the national peg. The only use of that forecast is a warning before a quarterly table review: a lane mix that relays through a persistently expensive region may be mis-charged by a national schedule, and finance may want to renegotiate. The forecast does not replace the contract schedule on a live invoice. It does not choose the peg, the increment, or the cents. It does not hedge a fuel purchase, and this page does not describe how anyone would trade.
The failure mode to design against is a model that learns the surcharge away. Trained on all-in paid amounts, it will treat last quarter's peg as if it were part of the lane rate. The next peg move then shows up as a linehaul error, or the billing step adds a full surcharge on top of a linehaul that already contains one. Keep the training target on linehaul alone. Keep the schedule in the rating step. A probability model and a rate model are both guests of the schedule. They do not edit it.
If the warning is wide and the contract is national, the human outcome is a conversation at renewal, not a silent regional rate. If the warning is narrow, the outcome is to leave the table alone. In both cases the invoices between reviews keep using the version in effect. A feasibility warning that leaks into the rating job has stopped being a warning. Treat that leak as an incident, the same as an unversioned edit to the base price.
Do
- Keep the linehaul training target on linehaul alone
- Keep the schedule in the rating step
- Bring a wide regional warning to the renewal conversation
Avoid
- Training on all-in paid amounts
- Swapping to a regional series the contract does not name
- Letting a forecast file update the peg table
Frequently asked questions
The questions below use the same words as the structured data on this page. They are the questions a billing owner should settle before a peg is applied without a clerk on every invoice.
The answers keep the surcharge on the contract schedule. They do not turn a forecast into a rate, and they do not describe a fuel trade.
What do freight contracts mean by the DOE fuel peg?
The DOE peg in these contracts is the EIA weekly retail on-highway diesel price, the national series unless the contract names a regional series. The surcharge is a deterministic table applied after linehaul: a base price, an increment, and cents per mile or a percent of linehaul. A forecast is not the peg.
Why must a neural model not learn the fuel surcharge?
If the linehaul model absorbs the surcharge, billing will add the peg a second time or hide it where an audit cannot replay the schedule. The model may forecast linehaul only. The peg stays a contract rule. Neural hedging on this page is only a feasibility warning for a quarterly table review.
How should fuel surcharge be split across legs without double counting?
Apply the peg once to the miles the contract actually bills. Split that one amount by each leg's share of those miles, so the parts add back to the total. Do not charge the full surcharge on every leg. Two separate contracts are two calculations, not a third charge on top.
What does the invoice audit compare the fuel line with?
The audit compares billed fuel surcharge with the contract schedule for the peg version in effect, the miles, and the rounding rule. It does not compare the invoice with a neural forecast. A difference beyond the penny tolerance goes to a biller. A forecast that disagrees with the schedule is not a discrepancy.
Who is allowed to change the peg table?
A named finance owner and the contract owner change the table together, with a new effective date. The model may not change it. A quarterly review may read a regional-basis warning and then decide whether to renegotiate. Invoices already rated keep the version they used.
Bill fuel once, from the schedule, every time.
Strong Tower Consulting provides independent systems architecture, feasibility audits, and engineering roadmaps for motor carriers and freight brokers. Contact our advisory team or call 763-244-1373 to discuss your operational data and evaluate practical implementation paths for your fleet.
Published October 23, 2026 by Strong Tower Consulting, LLC. Founded 2008 in Minnesota by Scott Elliott.