Typical Hard Money Loan Terms: Rates, Points and Leverage (2026)

Rates and points: the dated numbers

Three sources publish pricing compiled across many lenders' loans rather than one lender's rate sheet. They measure different things over different periods, so read each row with its date and its sample.

Medians are not averages, and samples move. SFR Analytics' figures come from data that lenders submit, and the set of lenders can change from month to month. Between the March and June 2026 reports the median points fell from 2.0 to 0.98. Treat that as a change in the reported numbers, not proof that the whole market cut its points in half.

The 2024 data shows the trend. In the AAPL article by Nema Daghbandan, drawing on Lightning Docs loan data, the national average bridge rate fell from 11.6% in January to 10.99% in October 2024, and bridge points averaged 2.09% that October, after running as high as 2.33%. The same article reports that 38% of bridge loans had no construction holdback, and that of 8,832 bridge construction loans, 28% charged Dutch interest and 72% charged non-Dutch. A headline rate on a Dutch loan costs more than the same rate charged only on drawn funds.

Educational ranges are wider. AAPL's July 2025 master class article on terminology, by Anthony Geraci, describes origination fees as "often 1-3 points," rates that "typically range from 7-15%" and a conservative LTV of "e.g., 65-70%." Those are teaching ranges, not survey results.

One point is 1% of the loan amount (the CFPB uses the same definition for mortgage discount points). For context on market share, the NPLA report says private lenders financed 16.4% of investor single-family purchases from July 1, 2025 to June 30, 2026.

What the 2026 medians mean on one loan

Rates and points only mean something in dollars. Here is the site's example deal, a $280,000 loan over 12 months with interest on the full balance, priced once at each 2026 median. Nothing else changes. This is an illustration of the arithmetic, not a quote.

Of the $4,284.00 difference over the full term, $2,856.00 is points. Points are collected in full at closing while the rate difference accrues month by month, so on an early payoff points make up an even larger share of the gap. The hard money points calculator compares two quotes month by month, and how much hard money lenders make explains why gross lender revenue isn't profit.

Leverage limits lenders publish

There is no industry survey of leverage limits in the sources we found. What exists are individual lenders' published programs and examples. The four below are examples, not standards. "Up to" marks a program maximum that a given deal may not reach, and each lender measures its ratios its own way.

New Silver's example is the most useful because it shows the arithmetic: a $200,000 purchase, a $50,000 rehab and a $290,000 ARV, sized at 90% LTC and 75% of ARV, with the lower amount winning and the rehab held back. Run through this site's engine with the same inputs, the maximum loan is $217,500 with LTARV binding (the LTC cap is $225,000). The holdback is $50,000 and the acquisition advance $167,500, matching the help-center figures; the loan's actual LTC works out to 87%.

Flynn's rule shows a different convention. He applies the as-is limit to the money funded at closing, then takes "the more conservative of the parameters." This site's calculators apply As-Is LTV to the total loan, holdback included, so on a deal with a holdback the same 80% can produce a smaller loan here than under his method. Easy Street's "90% of purchase, 100% of rehab" is a split structure. The calculators model the rehab half with a 100% rehab funding setting but have no separate cap on the purchase share, so check the acquisition advance against 90% of the price yourself. LTV vs LTC vs LTARV walks through all three meanings of "LTV."

Bank supervisory LTV limits (context only)

Regulated banks work under the Interagency Guidelines for Real Estate Lending Policies, published by the OCC, the FDIC and the Federal Reserve. The guidelines set the ceilings that an insured institution's internal LTV limits should not exceed. They are written for banks, not for private lenders, but they show where a regulator draws conservative lines.

* No limit is established, but at 90% LTV or more at origination the guidelines expect credit enhancement such as mortgage insurance or readily marketable collateral.

Two definitions in the guidelines are worth knowing. A construction loan includes one for "rehabilitating buildings," and for a loan to purchase an existing property, "value" means "the lesser of the actual acquisition cost or the estimate of value." Some private lenders measure LTV the same way. The guidelines also say loan disbursements "should not exceed actual development or construction outlays," which is the logic behind holding back rehab money until the work is done. Whether any rule applies to your lending depends on who you are and where you lend; ask counsel.

Why your terms will differ

A median describes the middle of someone else's loans. The quote a borrower gets, or the terms a lender should offer, turn on the specifics:

• Borrower. Credit, liquidity after closing, and the number of similar projects completed. Programs that publish leverage tiers can tie the top tier to experience.
• Property. Type, condition, location and how heavy the rehab is relative to the purchase.
• Market. How quickly comparable homes sell and how confident the lender is in the ARV.
• Leverage against price. A lender may offer more leverage at a higher rate or more points, or better pricing for a lower loan amount. Compare offers at the same loan amount.
• Structure. Dutch or non-Dutch interest, minimum interest, fees, and whether the as-is limit applies to the whole loan or only the initial advance. Two quotes with the same headline rate and LTV can differ by thousands of dollars.
• Timing. The figures on this page are dated. Rates and points move.

To test a specific deal, enter the lender's actual limits in the loan sizing calculator. It shows each cap side by side and which one binds.