CPAEX

ETF performance based on market prices. Drag on chart to select period. Sharpe Ratio uses the one-month T-bill as the risk-free rate. From 10/7/2025 to 10/7/2026.

ETF performance based on market prices. Returns for periods greater than one year are annualized. As of 10/7/2026.

* partial year — the series ran for only part of it (its first year, or the year it closed); the figure is the return over the part it did run. ETF performance based on market prices. As of 10/7/2026.

Family

Counterpoint Mutual Funds

Registrant

NORTHERN LIGHTS FUND TRUST III

Type

Mutual Fund

Asset Class

Other

Peer Asset Class

Asset Allocation

Region

US

Equity Sleeve Style

Mid Core

Prospectus Benchmark

SP500 Index TR1077 Index

Peer Group

Allocation - Tactical/Dynamic (94 peers)

1Y Peer Percentile

63

3Y Peer Percentile

85

Expense Ratio

1.87%

5Y Tax Drag

1.02%/yr

AUM

$763M

3Y Flow

+797.3%

Yield

0.32%

Inception

11/30/2015

Holdings (Reported)

864

Turnover

236%

Equity

58%

Derivatives

4%

Net Equity Exposure

67%

Option Notional

-269%

The fund invests at least 80% of its assets in long and short positions in equity instruments of any market capitalization, including domestic and foreign common stock, preferred stock, and equity derivatives. It employs a proprietary machine learning model for stock selection, targeting undervalued securities for long positions and overvalued ones for short positions, while also varying equity exposure based on market return forecasts. The remaining 0-20% of assets are allocated to U.S. treasury instruments and investment grade debt.

Expense ratio and its components from the fund's prospectus fee table; tax drag from its after-tax return table. Turnover from the SEC annual report (12/8/2025). Yield as of 10/6/2026, AUM 9/30/2026. Holdings, asset class, style, peer group and exposures as of 6/30/2026. Strategy summary from the fund's prospectus.

Equity size and style definitions: Large vs. Small based on market cap, where Large > $70B, Mid $10-70B, and Small < $10B. Value vs. Growth based on percentile ranking on Traditional Value score within the fund's region and size cohort (equal thirds among its true equity-fund peers). This fund is not an equity fund, so the box describes its equity sleeve only (58% of assets): the sleeve's holdings are scored and ranked against equity funds of the same region and size cohort. The rest of the portfolio (bonds, cash, other assets) is not represented here. Cell shading and percentages: share of classified holdings weight per box, from stock-level classification (size on the same $70B/$10B lines; value by Traditional Value terciles across scored stocks). The dot is the average holding ranked against other funds, so dot and weights can differ at the margin. The Size and Style bars are the grid's row and column totals. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

Hover or click for factor definitions. Weighted average of underlying company-level values with weights equal to position size. Ranked within the clean equity fund universe, one entry per fund series (lowest: 0, highest: 100; the middle of the range is denser than the ends). Blue dots mark the fund; the shaded band spans the Asset Allocation Funds class 25th–75th percentile of scores, notched at the median. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

Hover or click for factor definitions. Weighted average of underlying company-level values with weights equal to position size. Solid blue dots mark the fund's position within the clean equity fund universe (5th–95th percentile scale); the shaded band spans the fund's broad class (Asset Allocation Funds) 25th–75th percentile, notched at the median. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026. Expense ratio, yield, and turnover as of 10/6/2026. AUM as of 9/30/2026.

Shares of the fund's long book, from the asset category each position is filed under. Holdings that are themselves funds are looked through one level to their own mix (33% of the book). Cash & Short-Term includes 0.4% of net assets the filing does not itemise. The short book is stated under the card and never netted into a slice: a pie cannot draw a negative, and folding a hedge into the class it offsets would make it look like more of that class. Other is everything we could not name as an asset class: holdings the filing itself labels as other, derivative positions at the market value the filing reports rather than at notional, and funds held inside this fund whose own holdings we could not read.

Shares are percentages of the holdings that carry a country, not portfolio weights. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

Shares are percentages of the fund's classified holdings — every position we can identify, with cash and unidentified positions excluded — not portfolio weights. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

Shares are percentages of the fund's classified equity book, renormalized so each split sums to 100% (the equity book is 58% of net assets) — they are not portfolio weights. China exposure: domiciled = incorporated in China or Hong Kong, falling back to the filing's country for holdings outside the stock model. Among the rest, revenue = ≥15% of revenue derived from China; operations = ≥3% of LinkedIn-tracked workforce in China, or ≥2 identified Chinese suppliers making up ≥12% of those identified. China revenue is estimated where segment reporting is absent. Company size bands by holding market cap: Mega >$500B, Large $70–500B, Mid $10–70B, Small $1–10B, Micro <$1B. Old vs New Economy partitions Sparkline's 48 industries: New = technology, internet platforms and entertainment, e-commerce, payments, pharma/biotech/medtech, and renewable energy; Old = everything else. Shares are of classified holdings and sum to 100%. China shares are of total equity weight. Size shares are of scored holdings weight. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

307 positions (top 10 shown, create free account for full holdings) · 32.4% in funds

Sector, country and factor detail need a security match; the share of weight that matched is a member figure, so the rows without one appear here with those columns blank. "In funds" adds up the rows tagged Fund — holdings that are themselves funds this app covers, which is what look-through can expand; a held vehicle that files no holdings of its own (a commodity or currency trust, a money-market fund) is not counted and stays a single line. Holdings as of 6/30/2026. Stock-level data as of 8/31/2026.

Hover or click for definitions. Solid blue dots mark the fund: percentage rows on a 0–100% scale, holdings counts log-scaled across the clean-universe 5th–95th percentile range. The shaded band spans the fund's broad class (Asset Allocation Funds) 25th–75th percentile, notched at the median. Holdings as of 6/30/2026.

Sparkline's own holdings-based peer taxonomy (189 share classes across 94 funds). One share class shown per fund — the exchange-listed class where the SEC series has one, else its longest-history class; percentile ranks and medians are computed against the same deduped set. Quartile chips fill the layer the fund's rank falls in (top layer = best quartile). Funds without a full return history rank only on the horizons they cover.

Nearest funds across the whole universe, by holdings and behavior — the peer taxonomy is ignored, so matches can come from any category.

The method columns are members-only — create a free account to see these figures.

Rows are the consensus top matches (a fund must place in at least two methods' top 250 to qualify, and matches missing more than one of this fund's own methods are not ranked — a row scored on fewer methods is a noisier estimate, not a better match). Click a method column to see that method's own top matches across the whole universe instead. Gray (N) beside each method's metric is that fund's rank in the whole universe on that method. Similarity is a percentile among match-grade fund pairs, 0–100 — 87 means the pair is closer than 87% of the strongest matches in the universe, averaged across the methods — so it compares across funds: index clones tie at the top with thousands of other clone pairs, so scores top out in the high 80s to 90s, while a one-of-a-kind fund's best match can score in the 30s. Rows are ordered by the Consensus Rank — the geometric mean of those universe ranks, shown in each cell's tooltip — and the score is held down the list to the level of the closest match above it, so equal scores mean equally close and the number reads as a floor. A dash means that method has no figure for the pair at all (one of them lacks the holdings or return history it needs). Hover a column header for each method's definition. The “Top…” strips above compare against every fund's own top match across the whole universe (they deliberately don't follow the Peer Universe selector); Top Similarity is simply the best match's score, with no reference band; the Holdings and Active Tilt strips carry no reference band because computing every fund's best book overlap would require a full holdings scan per fund — the dot stands alone on its natural 0–100% scale. One share class shown per fund, and a fund's own sister share classes are excluded from its Top-match strips. Holdings as of 6/30/2026. Consensus Rank is the geometric mean of the match's universe-wide rank across the methods (#1.0 = top match on every one). Holdings Overlap = common-holdings weight (sum of the minimum weight across shared positions, long only). Active Tilt Overlap = the same statistic on active weights (holdings minus the fund's own default benchmark, floored at 0), which strips the benchmark ballast that dominates raw overlap. Factor Similarity = 100 − a signature-weighted RMS gap across the composite factor scores (0–100 ranks within the clean equity fund universe), each axis weighted by how far this fund sits from the universe median of 50. Return Correlation = correlation of market-hedged daily returns over the trailing two years (CAPM residuals), so only market-beta co-movement is removed; requires ~18 months of daily history (378 of the window's 504 trading days). Fama-French Similarity = 1/(1+d), where d is the Euclidean distance between factor betas (market, size, value, profitability, investment, momentum, intangible value) from up to three years of daily returns (min 378 daily observations and regression R² of at least 0.25); 1.00 = identical loadings.